清理历史雷达RTK资料并归档当前车辆结果

改为车头向前整链：主从装反机械初值、双天线 pitch/roll 姿态与默认 HeadingOffsetDeg=-90
Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-25 14:45:31 +08:00 · 2026-08-11 18:07:54 +08:00 · 2026-08-10 22:58:46 +08:00
40 changed files with 1107 additions and 4753 deletions
@@ -1,25 +1,26 @@
 # 双天线RTK—3D LiDAR直接手眼标定

-本仓库从静态站点原始数据复现 `T_RTK_lidar`：把原始雷达点变换到 **RTK 基线导航系**。  
-它**不是** `base_link` 车体外参；求解阶段不使用车体航向偏置，也不使用 RTK 到后轮轴的 XY 杆臂。
+本仓库从静态站点原始数据复现 `T_RTK_lidar`：把原始雷达点变换到 **车头向前的 RTK 车体系**（主天线原点）。  
+求解不使用 RTK 到后轮轴的 XY 杆臂；与雷达–IMU 外参对照时旋转系一致，平移仍差天线原点。

 当前交付标定（2026-08 室外车，27 站）约定如下：

 | 项 | 值 |
 |---|---|
-| RTK 坐标系 | **基线系**（`HeadingOffsetDeg = 0`） |
+| RTK 坐标系 | **车头向前**（`HeadingOffsetDeg = -90`；主从装反、基线朝右） |
 | 天线相位中心离地高 | **1.9165 m**（1916.5 mm） |
-| 机械平移初值 | `(0.414179474, 0.210859360, 0.004000001) m` |
-| 机械旋转初值 | yaw ≈ **90°**（雷达 X 朝车头、双天线基线左右装） |
+| 机械初值（车头系） | \(t=(+0.21086,-0.41418,+0.07850)\) m，yaw=**0°**（CAD 纵向已按车头正向取 +X） |
+| 物理基线 | `baseline_points=vehicle_right`（主天线车左，从天线车右，后轴中心左右对称） |
+| 姿态 | 双天线 pitch/roll：`R = Rz(yaw_raw) Ry(-pitch) Rx(roll) Rz(+90°)` |
 | 地面点 ROI | LiDAR 系 **`z ∈ [-2.5, -1.5]`**（约 2 m 车顶安装） |
 | pair 配准 | **禁止**使用外参 seed；B 与 X 独立 |

-数据下载（历史 data4 等）：https://fs.fairylandtech.com:5001/FRLD/#file_id=963272954246902180  
+数据下载：https://fs.fairylandtech.com:5001/FRLD/#file_id=966776353886090246  
 账号：lichun.qu@fairylandtech.com　密码：lichun.qu

 ---

-## 1. 输出坐标约定（基线系）
+## 1. 输出坐标约定（车头向前）

 统一约定 `T_A_B` 把 B 系点变换到 A 系：

@@ -27,22 +28,16 @@
 p_RTK = T_RTK_lidar · p_lidar
 ```

-本仓库默认 RTK 导航系（**基线系 / baseline_raw_heading**）：
+本仓库默认 RTK 导航系（**车头向前 / vehicle_forward_heading_offset**）：

- 原点：GGA 位置参考点（通常为 ANT1 相位中心，须结合接收机配置确认）；
- X 轴：`rawHeading` 双天线基线在水平面的投影；
+- 原点：GGA 位置参考点（主天线 / ANT1 相位中心）；
+- X 轴：车头向前（`rawHeading + HeadingOffsetDeg`，本车 `HeadingOffsetDeg = -90`）；
 - Y 轴：左；
 - Z 轴：上；
- ENU 航向：`yaw = 90° - rawHeading`（`heading_offset = 0`）；
- roll、pitch：轨迹中固定为 0。
+- 姿态：先在基线系应用双天线 pitch/roll，再乘固定 `Rz(-heading_offset)`；不是 IMU 融合姿态。

-> 不要把基线系结果当成“车头向前系”。若下游需要车头向前，应另乘确认过的固定航向偏置，或显式使用 `-HeadingOffsetDeg 90` **整链重跑**，不要事后只改 JSON 里的 yaw。
-
-若下游需要 `T_body_lidar`，须另有已确认的 `T_body_rtk`：
-
-```text
-T_body_lidar = T_body_rtk · T_RTK_lidar
-```
+> 改 `HeadingOffsetDeg` 或姿态模型后必须从 **prepare** 起重跑；禁止事后只改 JSON 里的 yaw。  
+> 旧基线系结果（`HeadingOffsetDeg = 0`）与车头系外参不可混用。

 机械初值文件：[`run/rtk_lidar_mechanical_initial.json`](run/rtk_lidar_mechanical_initial.json)  
 **仅用于 AX=XB 求解初值，禁止用于 LiDAR pair 配准。**
@@ -54,7 +49,7 @@ T_body_lidar = T_body_rtk · T_RTK_lidar
 ```text
 原始雷达 + RTK（+ 可选 IMU）
  → combined/（按站关联的多传感器 NPZ）
-  → 每站选一帧静态点云 + yaw-only RTK pose（基线系）
+  → 每站选一帧静态点云 + RTK pose（车头向前，含双天线 pitch/roll）
  → Open3D GICP 与 small_gicp 分别求 B_ij = T_Li_Lj（无外参 seed）
  → 留出点、正反向、旋转共轭不变量等精筛
  → 双后端共识边 → consensus B
@@ -137,10 +132,10 @@ $Data = "D:\data\rtk_lidar_run"   # 含 combined/

 powershell.exe -NoProfile -ExecutionPolicy Bypass -File "$Repo\run\run_direct_rtk_lidar.ps1" `
  -CombinedRoot "$Data\combined" `
-  -WorkRoot "$Data\prepared_baseline_h19165" `
-  -OutputRoot "$Data\outputs_baseline_h19165" `
+  -WorkRoot "$Data\prepared_vehicle_h19165" `
+  -OutputRoot "$Data\outputs_vehicle_h19165" `
  -RtkReferenceHeightAboveGroundM 1.9165 `
-  -HeadingOffsetDeg 0 `
+  -HeadingOffsetDeg -90 `
  -ExpectedStations 27 `
  -MinStations 20 `
  -GroundZMin -2.5 `
@@ -153,7 +148,7 @@ powershell.exe -NoProfile -ExecutionPolicy Bypass -File "$Repo\run\run_direct_rt
 | 参数 | 本次取值 | 说明 |
 |---|---|---|
 | `-RtkReferenceHeightAboveGroundM` | **1.9165** | GGA/ANT1 相位中心离地高（m），必填 |
-| `-HeadingOffsetDeg` | **0** | 基线系；非 0 时才变成车头向前系 |
+| `-HeadingOffsetDeg` | **-90** | 车头向前（主从装反、基线朝右）；`0` 才是基线系 |
 | `-GroundZMin/Max` | **-2.5 / -1.5** | 约 2 m 车顶雷达；旧默认 `[-1.4,-0.4]` 会拟合到墙 |
 | `-ExpectedStations` | **27** | 本批站数 |
 | `-MinStations` | **20** | 远程旧脚本曾写死 30，会跑不了本批 |
@@ -188,17 +183,6 @@ $Out/
    └── final_T_RTK_lidar.json
 ```

-### 5.3 远程旧一键复现为何不能直接套用本批
-
-相对当前本地默认，远程 `origin/feature/lidar-rtk-direct-calibration` 仍有几处与本次标定不符：
-
-1. 地面 ROI 落到 py 默认 `[-1.4, -0.4]`（本车会拟合墙面）；
-2. `run_direct` 里 `MinStations` 曾写死 30（本批 27 站失败）；
-3. 缺少 `export_g90_h32_windows_to_combined.py`（本批导出链路）；
-4. README 示例高度仍写历史车 **0.758 m**（本车应为 **1.9165 m**）。
-
-航向上远程已是 `HeadingOffsetDeg 0`（基线系），与本次坐标系一致；请用**本分支本地提交**复现，不要照抄未更新的远程文档数字。
-
 ---

 ## 6. 3D 可视化
@@ -207,8 +191,8 @@ $Out/

 ```powershell
 $Repo = "D:\First-dev-dept\calibration-rtk-run"
-$Out  = "D:\data\rtk_lidar_run\outputs_baseline_h19165"
-$Work = "D:\data\rtk_lidar_run\prepared_baseline_h19165"
+$Out  = "D:\data\rtk_lidar_run\outputs_vehicle_h19165"
+$Work = "D:\data\rtk_lidar_run\prepared_vehicle_h19165"

 powershell.exe -NoProfile -ExecutionPolicy Bypass -File "$Repo\run\view_result.ps1" `
  -Frames "$Work\frames_all" `
@@ -241,32 +225,32 @@ powershell.exe -NoProfile -ExecutionPolicy Bypass -File "$Repo\run\view_result.p

 ---

-## 7. 当前标定结果（基线系，h = 1.9165 m）
+## 7. 当前标定结果（车头向前，h = 1.9165 m）

-结果目录：`D:\data\rtk_lidar_run\outputs_baseline_h19165\`  
-交付文件：`final_T_RTK_lidar.json` / `summary.json`
+> **状态：可作车头系候选交付**（`recommended_for_deployment: true`）。  
+> 约定：`HeadingOffsetDeg=-90`，双天线 pitch/roll，机械初值 \(t=(+0.21086,-0.41418,+0.07850)\)，yaw=0。  
+
+仓库内结果：[`results/vehicle_20260808/`](results/vehicle_20260808/)（来自本机 `outputs_vehicle_h19165`）。

 ```text
-translation_m = [0.412305582, 0.217309210, 0.104057606]
-RPY_deg_xyz   = [0.465513, 0.743343, 89.460018]
+translation_m = [0.217822250, -0.411347802, 0.106542337]
+RPY_deg_xyz   = [0.066239, 0.809662, -0.551322]

 T_RTK_lidar ≈
- 0.009424  -0.999922   0.008247   0.412306
- 0.999871   0.009529   0.012896   0.217309
-0.012973   0.008124   0.999883   0.104058
+ 0.999854   0.009639   0.014119   0.217822
+-0.009621   0.999953  -0.001292  -0.411348
+-0.014131   0.001156   0.999899   0.106542
 0          0          0          1
 ```

 | 指标 | 值 |
 |---|---:|
 | 有效站点 / 共识对 | 27 / 20 |
-| 平移残差 RMS / 中位 / P95 / max | 0.070 / 0.042 / 0.121 / **0.186** m |
-| 旋转残差 RMS / 中位 / max | 0.978 / 0.585 / **2.73** ° |
-| 双后端差 | 3.2 mm / 0.17° |
-| Jacobian 条件数 | 6.88 |
-| bootstrap σ(x,y,z) | 4.7 / 6.3 / 0.8 mm |
-| 相对机械初值 | 旋转差 ≈ 1.03°（无近 180° 冲突） |
-| `frame_mode` | `baseline_raw_heading` |
+| 平移残差 RMS | ≈ 0.071 m |
+| 旋转残差 RMS | ≈ 0.982 ° |
+| 双后端差 | ≈ 3.1 mm / 0.12° |
+| `frame_mode` | `vehicle_forward_heading_offset` |
+| 相对机械初值 | XY 近机械杆臂；yaw≈0；无近 180° 冲突 |

 与机械平移初值 XY 相差约数毫米；z 由天线高度约束，CAD 的 4 mm 不能代替实测 1.9165 m。

@@ -286,27 +270,22 @@ T_RTK_lidar ≈
 平面阿克曼运动不能独立观测 z。z 由「LiDAR 地面平面 + 外供 RTK 参考点离地高」约束：

 - 本次：**1.9165 m**（相位中心离地）；
- 历史 data4/data5 文档中的 **0.758 m** 是**另一台车**的测量，不能用于本车。
+- 不得复用其他车辆或历史采集的天线离地高度。

 更改高度后必须重新求解，禁止只改 JSON 里的 z。  
 GGA 对应哪根天线、`rawHeading` 方向须现场确认；搞反会导致 yaw 差约 180°。

 ---

-## 9. 历史 data4 / data4+data5（参考）

-旧联合实验使用 ANT1 离地 `0.758 m`，外参量级与本车不同，**不要与第 7 节结果混比**。联合流程见 `run/run_joint_rtk_lidar.ps1`。  
-`results/reference_data4/` 若存在，仅为历史精简产物，不作为当前交付外参。
-
---
-
-## 10. 仓库目录
+## 9. 仓库目录

 | 目录 | 职责 |
 |---|---|
 | [`code/`](code/) | GICP、运动对质量、AX=XB、结果封装、3D 可视化 |
 | [`tools/`](tools/) | dlog/rscap 解析、G90 窗导出、combined / prepared |
 | [`run/`](run/) | PowerShell 入口；路径与高度均由参数传入 |
+| [`results/`](results/) | 当前车辆的最终外参与质量摘要；不含原始数据和中间点云 |
 | `tests/` | 坐标契约、G90 host 关联等回归 |
 | `work/`、`outputs/` | 本地生成物（`.gitignore`） |

@@ -1,41 +0,0 @@
-{
-	"class_name" : "PinholeCameraParameters",
-	"extrinsic" : 
-	[
-		0.99504759165761958,
-		-0.069911734310665122,
-		0.070658614068252801,
-		0.0,
-		-0.09933361974813136,
-		-0.67348569005019454,
-		0.73249563635925308,
-		0.0,
-		-0.0036224748591220518,
-		-0.7358867947607759,
-		-0.67709489953226409,
-		0.0,
-		2.6365670299167046,
-		1.9293765342813529,
-		11.882725892422524,
-		1.0
-	],
-	"intrinsic" : 
-	{
-		"height" : 900,
-		"intrinsic_matrix" : 
-		[
-			779.4228634059948,
-			0.0,
-			0.0,
-			0.0,
-			779.4228634059948,
-			0.0,
-			699.5,
-			449.5,
-			1.0
-		],
-		"width" : 1400
-	},
-	"version_major" : 1,
-	"version_minor" : 0
-}
@@ -1,41 +0,0 @@
-{
-	"class_name" : "PinholeCameraParameters",
-	"extrinsic" : 
-	[
-		0.99504759165761958,
-		-0.069911734310665122,
-		0.070658614068252801,
-		0.0,
-		-0.09933361974813136,
-		-0.67348569005019454,
-		0.73249563635925308,
-		0.0,
-		-0.0036224748591220518,
-		-0.7358867947607759,
-		-0.67709489953226409,
-		0.0,
-		2.6365670299167046,
-		1.9293765342813529,
-		11.882725892422524,
-		1.0
-	],
-	"intrinsic" : 
-	{
-		"height" : 900,
-		"intrinsic_matrix" : 
-		[
-			779.4228634059948,
-			0.0,
-			0.0,
-			0.0,
-			779.4228634059948,
-			0.0,
-			699.5,
-			449.5,
-			1.0
-		],
-		"width" : 1400
-	},
-	"version_major" : 1,
-	"version_minor" : 0
-}
@@ -1,240 +0,0 @@
-LiDAR、双天线 RTK、IMU 标定数据说明
-====================================
-
-文档日期：2026-07-23
-配套代码仓库：calibration
-标定目标：求解 3D LiDAR 到后轮轴中心车体系的外参 T_body_lidar。
-
-
-一、重要说明
------------
-
-1. 三批数据的记录格式和用途并不完全相同。export 是每站多帧的完整解码数据，prepared 是从中每站选一帧并配好 RTK 车体位姿后的标定输入。
-2. 第一批（data1）、第二批(data2)目前上传的是已经从原始 dlog 数据以及导出的 NPZ 数据。
-3. data4 保留了完整逐站 LiDAR dlog 和独立 RTK/IMU rscap，可以从原始记录开始复现。
-4. 每一个站点采集 LiDAR 点云时车辆均静止，因此当前 LiDAR-RTK 标定没有使用 IMU进行点云运动畸变校正。
-5. data4 中的 IMU 数据保持在 IMU 原始传感器坐标系；当前流程只解析、关联和保存 IMU，没有求解 IMU 外参。
-6. 不要修改站点目录名称、prepared 中 station_*.npz 的顺序或 manifest。B 文件中的运动对索引依赖这些顺序。
-
-
-二、目录结构
----------------------
-
-LiDAR_RTK_calibration_data/
-  数据说明.txt
-  data1/
-     raw/
-    export/
-    prepared/   
-  data2/
-      raw/
-    export/
-    prepared/
-  data4/
-    raw/
-      RTKIMUraw/
-    prepared/              
-    combined/(三传感器混合后的数据，data1、2没有IMU的数据。)      
-说明：
-
- data1/export 和 data2/export 是从逐站 dlog 导出的数据。raw 是完整原始数据。
- data1/prepared 和 data2/prepared 是每站选择一帧并重建 RTK 车体位姿后的标定输入。
- data4/raw 是完整原始数据，能够重新生成 export、parsed、combined 和 prepared。
-它们可以由 raw 重新生成。
-
-
-三、第一批数据（data1）
-----------------------
-
-1. 数据范围
-
- 静止站点数量：38 站。
- 站点编号：01～38。
- 传感器：3D LiDAR + 双天线 RTK。
- 不包含 IMU。
- 记录格式：旧式逐站 dlog 中同时记录 LiDAR 和 GPS-POST-Z；当前云盘包从已导出的 NPZ 开始。
-
-2. RTK 特点
-
- 实际 RTK 记录约 10 秒一条，并非预期的 10 Hz。
- 每站有效 RTK 样本较少，部分站点约 1～11 个有效样本。
- 因车辆在每站静止，仍可对站内 RTK 样本做均值并构造站点位姿；但时间同步精度和航向统计能力弱于第二批。
-
-3. 当前用途
-
- 第一批只作为辅助复核数据。
- 不作为当前部署外参的主要求解数据。
- 不应把第一批写成严格独立的“验证集”，因为 RTK 过于稀疏，且它和第二批的场景、采集流程相近。
-
-4. 上传内容
-
-data1/export/
-
- 当前文件数约 803。
- 当前大小约 0.088 GB（约 90 MB）。
- 含逐站导出的 LiDAR NPZ、RTK sidecar、manifest 和质量报告。
-
-data1/prepared/
-
- 当前文件数约 118。
- 当前大小约 0.049 GB（约 50 MB）。
- 核心内容包括：
-  frames_all/station_*.npz
-  body_poses_rear_gga_raw_rear_to_front.csv
-  station_summary.csv
-  manifest.json
-
-
-四、第二批数据（data2）
-----------------------
-
-1. 数据范围
-
- 静止站点数量：38 站。
- 原始站点编号：39～76。
- prepared 中重新顺序编号为 station_01～station_38。
- 传感器：3D LiDAR + 双天线 RTK。
- 不包含 IMU。
- 记录格式与第一批相同，但 RTK 采样正常且明显更密集。
-
-2. RTK 特点
-
- 每站约有 125～412 个有效 RTK 样本。
- 已使用 fix 4/fix 5 和有效 heading 进行筛选。
- 站内 heading 圆标准差上限使用 0.5°，本批 38 站均通过。
-
-3. 当前用途
-
- 第二批是当前部署外参的主要求解数据。
- 使用 small_gicp 和 Open3D GICP 分别求 B，再做与 X 无关的质量筛选和跨后端一致性筛选。
- 当前部署外参主要由第二批求得，第一批仅辅助复核。
-
-4. 上传内容
-data2/export/
-
- 当前文件数约 759。
- 当前大小约 0.127 GB（约 130 MB）。
- 含逐站导出的 LiDAR NPZ、RTK 信息、manifest 和质量报告。
-
-data2/prepared/
-
- 当前文件数约 82。
- 当前大小约 0.033 GB（约 34 MB）。
- 核心内容包括：
-  frames_all/station_*.npz
-  body_poses_rear_gga_raw_rear_to_front.csv
-  station_summary.csv
-  manifest.json
-
-
-五、data4 数据
--------------
-
-1. 数据范围
-
- 静止站点数量：34 站。
- 站点编号：001～034。
- 传感器：3D LiDAR + 双天线 RTK + IMU。
- LiDAR 位于每个站点自己的原始 dlog 中。
- RTK 和 IMU 位于独立 rscap 文件中，存放在 RTKIMUraw 目录。
- 原始目录共约 152 个文件，大小约 12.488 GB。
-
-2. RTK/IMU 原始记录
-
- RTKIMUraw 中当前包含 3 个 RTK rscap 和 3 个 IMU rscap。
- 本次 34 站标定使用 20260723-051627 开始的长时间 RTK/IMU session。
- 该 session 的 capture 审计结果：
-  RTK：54256 个记录块，missing_chunks=0，bad_record_crc=0，干净关闭，footer CRC 有效。
-  IMU：26719 个记录块，missing_chunks=0，bad_record_crc=0，干净关闭，footer CRC 有效。
-
-3. 时间关联与导出结果
-
-处理顺序为：
-
-统一时间轴
-  -> 分别解析 LiDAR、RTK、IMU
-  -> 按每个 LiDAR 帧关联最近有效 RTK
-  -> 保存 LiDAR 帧前后各 100 ms 的 IMU 窗口
-  -> 导出 combined NPZ
-  -> 每站选择一个静止帧生成 prepared
-
-当前关联统计：
-
- LiDAR 帧总数：11678。
- 34 个站点全部有数据。
- rtk_valid：11678。
- heading_valid：11678。
- fixed RTK：11678。
- IMU 窗口非空：11678。
- RTK 最大允许关联时间差：150 ms。
- IMU 窗口：LiDAR 时刻前后各 100 ms。
-
-时间基础：LiDAR 和串口 host UTC 用于当前关联；RTK GNSS 时间和 IMU 设备时间同时保留，供后续进一步建立精确时钟模型。
-
-4. 当前用途
-
- data4 用于独立重新求解一套外参，并与历史第二批结果做跨批比较。
- data4 中 IMU 没有参与当前 LiDAR-RTK 外参求解。
- data4 结果与历史部署外参相差约 1.592 cm / 0.234°，但 data4 自身 AX 残差更高，因此当前仍保留历史第二批结果作为部署值。
-
-如果已经上传 data4/raw，则 export、combined 和 parsed 均可以用代码重新生成。为了节省云盘空间，可只额外上传 prepared 和 calibration。
-
-
-六、三批数据差异汇总
--------------------
-
-第一批：
- 38 站，旧式 LiDAR+RTK dlog，无 IMU。
- RTK 极稀疏，约 10 秒一条。
- 当前上传从 export 开始。
- 只用于辅助复核。
-
-第二批：
- 38 站，旧式 LiDAR+RTK dlog，无 IMU。
- RTK 密集、航向稳定。
- 当前上传从 export 开始。
- 用于当前部署外参的主要求解。
-
-data4：
- 34 站，逐站 LiDAR dlog + 独立 RTK/IMU rscap。
- 保存完整原始数据，可从 raw 开始复现。
- 用于独立重算和跨批比较。
- IMU 只保存和关联，尚未完成 IMU 外参标定。
-
-
-七、标定坐标与主要参数
----------------------
-
- 外参定义：T_body_lidar，将 LiDAR 原始点变换到后轮轴中心车体系。
- 车体系：x 向前，y 向左，z 向上。
- 手眼方程：A_ij X = X B_ij。
- A_ij：由 RTK 后轮轴中心位置和双天线 heading 构造；当前为 yaw-only 姿态。
- B_ij：由两个静止站点的原始 LiDAR 点云通过 GICP 求得。
- heading_offset_deg：21.226°（本车安装参数，不是通用常数）。
- 后天线在车体系杆臂：[ -0.320, -0.365, 0.620 ] m。（手量）
- 后轮轴中心离地高度：0.2335 m，用于地面约束；不是 LiDAR 离地高度。
-
-
-八、数据使用注意事项
--------------------
-
-1. data1、2 的manifest 中可能仍保留旧脚本生成的 train/validation 字段。这些字段是历史元数据；当前严谨流程将第二批用于求解、第一批用于辅助复核，不把同一批内部的小样本划分描述为高可信度验证集。
-2. 不要使用已经变换到车体系的点云求 B，必须使用 NPZ 中的 points_raw。
-3. 可视化时，frames_all 必须与生成 B 文件时的站点数量和顺序完全一致。
-4. 模式 3（GICP B）本身已经错位时，应优先检查点云配准和场景退化；只有模式 3 正常而模式 4（X^-1 A X）系统性错位时，才优先检查 RTK A、坐标约定或外参 X。
-5. 当前结果是工程标定结果，不是由全站仪或高精度标靶认证的绝对真值。
-
-
-九、配套代码位置
----------------
-
-本机代码仓库：
-calibration
-
-根 README.md 包含：
- 从原始数据/导出数据开始的完整复现流程；
- code、tools、run 中每个主要文件的职责；
- small_gicp、Open3D GICP、consensus B 的处理逻辑；
- AX=XB 与地面约束求 X 的方式；
- 残差、Hessian/条件数、bootstrap、跨批检查和 3D 可视化方法。
-
@@ -15,7 +15,17 @@ def load(path: Path) -> dict:

 def write(path: Path, document: dict) -> None:
    path.parent.mkdir(parents=True, exist_ok=True)
-    path.write_text(json.dumps(document, ensure_ascii=False, indent=2), encoding="utf-8")
+
+    def default(obj):
+        if isinstance(obj, (np.bool_, np.integer)):
+            return obj.item()
+        if isinstance(obj, np.floating):
+            return float(obj)
+        if isinstance(obj, np.ndarray):
+            return obj.tolist()
+        raise TypeError(f"Object of type {type(obj).__name__} is not JSON serializable")
+
+    path.write_text(json.dumps(document, ensure_ascii=False, indent=2, default=default), encoding="utf-8")


 def inverse(t: np.ndarray) -> np.ndarray:
@@ -34,12 +44,127 @@ def delta(a: np.ndarray, b: np.ndarray) -> dict:
    }


+def wrap180(deg: float) -> float:
+    return (deg + 180.0) % 360.0 - 180.0
+
+
+def yaw_deg_of(transform: np.ndarray) -> float:
+    return float(Rotation.from_matrix(transform[:3, :3]).as_euler("xyz", degrees=True)[2])
+
+
+def mechanical_self_consistency(document: dict) -> dict:
+    """Reject mechanical JSON that mixes incompatible baseline / body definitions."""
+    translation = np.asarray(document["translation_m"], float)
+    yaw = float(document["rotation_rpy_deg_xyz"][2])
+    side = str(document.get("baseline_points", "")).strip().lower()
+    frame_mode = str(document.get("frame_mode", "")).strip().lower()
+    heading_offset = float(document.get("heading_offset_deg", 0.0) or 0.0)
+    vehicle_forward = (
+        frame_mode == "vehicle_forward_heading_offset"
+        or abs(heading_offset) > 1e-6
+    )
+    issues: list[str] = []
+
+    if vehicle_forward:
+        if abs(wrap180(yaw)) > 15.0:
+            issues.append(
+                f"vehicle-forward mechanical initial requires yaw≈0°, got {yaw:g}°"
+            )
+        lever = document.get("vehicle_flu_lever_master_to_lidar_m")
+        if lever is not None:
+            if float(np.linalg.norm(translation - np.asarray(lever, float))) > 0.05:
+                issues.append(
+                    "vehicle-forward translation_m must match vehicle_flu_lever_master_to_lidar_m"
+                )
+        if abs(heading_offset + 90.0) > 1e-6 and abs(heading_offset - 90.0) > 1e-6:
+            issues.append(
+                f"vehicle-forward heading_offset_deg should be ±90 for left/right baseline, got {heading_offset:g}"
+            )
+    elif side in {"vehicle_left", "left"}:
+        if abs(wrap180(yaw - (-90.0))) > 15.0:
+            issues.append(
+                f"baseline_points=vehicle_left requires yaw≈-90°, got {yaw:g}°"
+            )
+        if translation[0] <= 0.0 or translation[1] <= 0.0:
+            issues.append(
+                "baseline_points=vehicle_left expects +X/+Y lever in RTK baseline frame "
+                f"(got t_xy=({translation[0]:g}, {translation[1]:g}))"
+            )
+    elif side in {"vehicle_right", "right"}:
+        if abs(wrap180(yaw - 90.0)) > 15.0:
+            issues.append(
+                f"baseline_points=vehicle_right requires yaw≈+90°, got {yaw:g}°"
+            )
+        # Swapped but centerline-symmetric master (vehicle left): +X / -Y in baseline frame.
+        if translation[0] <= 0.0 or translation[1] >= 0.0:
+            issues.append(
+                "baseline_points=vehicle_right (master on vehicle left, baseline to the right) "
+                "expects +X/-Y lever in RTK baseline frame "
+                f"(got t_xy=({translation[0]:g}, {translation[1]:g}))"
+            )
+    else:
+        left_xy = translation[0] > 0.05 and translation[1] > 0.05
+        right_xy = translation[0] < -0.05 and translation[1] < -0.05
+        swapped_right_xy = translation[0] > 0.05 and translation[1] < -0.05
+        if left_xy and abs(wrap180(yaw - 90.0)) <= 15.0:
+            issues.append(
+                "mixed baseline definition: +X/+Y translation (left-baseline) combined with yaw≈+90° (right-baseline)"
+            )
+        if right_xy and abs(wrap180(yaw - (-90.0))) <= 15.0:
+            issues.append(
+                "mixed baseline definition: -X/-Y translation combined with yaw≈-90°"
+            )
+        if swapped_right_xy and abs(wrap180(yaw - (-90.0))) <= 15.0:
+            issues.append(
+                "mixed baseline definition: +X/-Y translation (swapped-master right-baseline) "
+                "combined with yaw≈-90° (left-baseline)"
+            )
+
+    return {
+        "baseline_points": side or None,
+        "frame_mode": frame_mode or None,
+        "heading_offset_deg": heading_offset,
+        "consistent": not issues,
+        "issues": issues,
+    }
+
+
+def solution_matches_declared_side(solution: np.ndarray, document: dict) -> dict:
+    """Check whether the solved extrinsic agrees with the mechanical baseline side."""
+    side = str(document.get("baseline_points", "")).strip().lower()
+    yaw = yaw_deg_of(solution)
+    t = solution[:3, 3]
+    expected_yaw = float(document["rotation_rpy_deg_xyz"][2])
+    yaw_err = abs(wrap180(yaw - expected_yaw))
+    xy_err = float(np.linalg.norm(t[:2] - np.asarray(document["translation_m"][:2], float)))
+    z_err = float(abs(t[2] - float(document["translation_m"][2])))
+
+    opposite_yaw = abs(wrap180(yaw - expected_yaw) - 180.0) <= 15.0 or abs(
+        wrap180(yaw - expected_yaw) + 180.0
+    ) <= 15.0
+    # Same XY sign as mechanical but yaw flipped ~180° (classic mixed inheritance).
+    same_xy_sign = (t[0] * float(document["translation_m"][0]) > 0.0) and (
+        t[1] * float(document["translation_m"][1]) > 0.0
+    )
+    mixed_inheritance = same_xy_sign and opposite_yaw
+
+    return {
+        "baseline_points": side or None,
+        "solution_yaw_deg": yaw,
+        "expected_yaw_deg": expected_yaw,
+        "yaw_error_deg": yaw_err,
+        "xy_error_m": xy_err,
+        "z_error_m": z_err,
+        "mixed_translation_rotation_inheritance": bool(mixed_inheritance),
+        "near_expected_pose": bool(yaw_err <= 15.0 and xy_err <= 0.25),
+    }
+
+
 def coordinate_contract_audit(raw: dict) -> dict:
-    """Compare the data-driven solution with the declared mechanical initial.
+    """Audit mechanical self-consistency and solution agreement.

    A near-180-degree disagreement is not auto-corrected: it normally means
-    that one physical forward-axis statement is reversed. Silently rotating
-    the point cloud would preserve residuals while changing the frame contract.
+    that one physical forward-axis / baseline-direction statement is reversed.
    """
    path_text = raw.get("solver_initial_extrinsic")
    if not path_text:
@@ -59,14 +184,29 @@ def coordinate_contract_audit(raw: dict) -> dict:
    solution = np.asarray(raw["matrix_4x4"], float)
    comparison = delta(initial, solution)
    near_180 = abs(comparison["rotation_deg"] - 180.0) <= 15.0
+    mech_check = mechanical_self_consistency(initial_document)
+    match = solution_matches_declared_side(solution, initial_document)
+
+    if not mech_check["consistent"]:
+        status = "mechanical_initial_inconsistent"
+    elif match["mixed_translation_rotation_inheritance"] or near_180:
+        status = "near_180_degree_axis_conflict"
+    elif not match["near_expected_pose"]:
+        status = "solution_disagrees_with_mechanical_baseline_side"
+    else:
+        status = "no_near_180_degree_axis_conflict"
+
+    requires = status != "no_near_180_degree_axis_conflict"
    return {
-        "status": "near_180_degree_axis_conflict" if near_180 else "no_near_180_degree_axis_conflict",
-        "requires_physical_axis_confirmation": near_180,
+        "status": status,
+        "requires_physical_axis_confirmation": requires,
        "mechanical_initial_path": str(path.resolve()),
+        "mechanical_self_consistency": mech_check,
+        "solution_vs_declared_baseline_side": match,
        "solution_relative_to_mechanical_initial": comparison,
        "note": (
-            "No automatic 180-degree point-cloud flip was applied. Confirm the Helios "
-            "aviation-connector side and the G90 vehicle-forward definition before deployment."
+            "No automatic 180-degree correction was applied. Confirm static GNHPR "
+            "left/right vs vehicle heading and Helios +X vs vehicle forward before deployment."
        ),
    }

@@ -87,12 +227,18 @@ def corrected(raw: dict, backend: str, reference_height: float, heading_offset_d
            "RTK": {
                "origin": "GGA positioning reference point; confirm ANT1/reference antenna in receiver configuration",
                "x_axis": x_axis,
-                "y_axis": "left",
+                "y_axis": "left of the RTK X/baseline axis (not necessarily vehicle-left)",
                "z_axis": "up",
                "yaw_enu_deg": f"90 - (rawHeadingDeg + {heading_offset_deg:g})",
                "frame_mode": "baseline_raw_heading" if baseline_frame else "vehicle_forward_heading_offset",
            },
-            "LiDAR": "raw LiDAR sensor frame",
+            "LiDAR": {
+                "description": "raw Helios sensor frame from points_raw polar decode",
+                "x_axis": "+X at azimuth 0° (forward when aviation connector faces vehicle rear)",
+                "y_axis": "+Y at azimuth +90° (left when +X is vehicle-forward)",
+                "z_axis": "up",
+                "origin_note": "optical/center per Helios manual; mounting height includes 63.5 mm base offset when deriving mechanical ΔZ",
+            },
        },
        "backend": backend,
        "measured_lidar_extrinsic_used_as_initial": bool(raw.get("measured_extrinsic_used_as_initial")),
@@ -154,18 +300,33 @@ def main() -> None:
    needs_axis_confirmation = bool(
        final["coordinate_contract_audit"]["requires_physical_axis_confirmation"]
    )
+    status = final["coordinate_contract_audit"]["status"]
+    reason_map = {
+        "mechanical_initial_inconsistent": (
+            "Mechanical initial mixes incompatible baseline-left/right translation and yaw; "
+            "fix run/rtk_lidar_mechanical_initial.json before trusting deployment"
+        ),
+        "near_180_degree_axis_conflict": (
+            "Physical axis confirmation is required because the data-driven solution differs "
+            "from the declared mechanical initial by approximately 180 degrees "
+            "(or inherits mixed translation/rotation signs)"
+        ),
+        "solution_disagrees_with_mechanical_baseline_side": (
+            "Solution yaw/XY disagree with the declared mechanical baseline side; "
+            "confirm static GNHPR direction before deployment"
+        ),
+    }
    final["selection"] = {
        "recommended": not needs_axis_confirmation,
        "reason": (
-            "Physical axis confirmation is required because the data-driven solution differs "
-            "from the declared mechanical initial by approximately 180 degrees"
-            if needs_axis_confirmation else
-            "Uses only motion pairs accepted independently by both Open3D GICP and small_gicp"
+            reason_map.get(
+                status,
+                "Uses only motion pairs accepted independently by both Open3D GICP and small_gicp",
+            )
        ),
        "open3d_vs_small_gicp": delta(open_t, small_t),
    }

-
    write(args.result_root / "final_T_RTK_lidar.json", final)
    summary = {
        "final": {
@@ -1,13 +0,0 @@
-# 数据说明
-
-原始 H32 dlog / G90·N300 `.rscap`（以及旧版 LiDAR dlog / `h32.rscap`）、逐帧 NPZ 和 prepared 点云体积较大，不进入 Git。请从项目云盘取得数据，并按根 README 中的目录示例放置；实际路径通过命令参数传入。
-
-推荐原始布局：
-
-```text
-raw_dataset/
-├── stations/<站号>/          # dobject/ + dobject_recording/（MSOP+DIFOP）
-└── captures/rtk.rscap, imu.rscap
-```
-
-公开数据包应同时提供：采集日期、车辆/传感器安装版本、站点数量、ANT1/ANT2 接线、rawHeading 方向、RTK 参考点离地高度及其测量方法。
@@ -1,5 +1,15 @@
-# results目录
+# 当前车辆标定结果

-`reference_data4/`是本仓库附带的精简参考结果。新的运行结果应写到仓库外目录或`outputs/`，不要覆盖参考结果。
+本目录只保存当前车辆、当前传感器安装条件下的最终可交付结果；不保存历史车辆数据、原始采集包、点云帧或中间配准产物。

-参考结果保留最终矩阵、共识B、两后端精筛B、逐对CSV/筛选审计和地面平面；未保留原始点云、逐帧combined数据、冗长的初筛JSON和带本机绝对路径的过程文件。
+## 2026-08 车辆 / 27 个静止站点
+
+目录 [`vehicle_20260808/`](vehicle_20260808/) 对应本机运行目录 `D:\data\rtk_lidar_run\outputs_vehicle_h19165`。
+
+- 外参：`final_T_RTK_lidar.json`
+- 质量摘要：`summary.json`
+- 坐标约定：`p_RTK = T_RTK_lidar · p_lidar`，RTK 为车头向前坐标系（`HeadingOffsetDeg=-90`）。
+- RTK 参考点高度：1.9165 m（ANT1 相位中心）。
+- 质量：27 个站点、20 个共识运动对、平移 RMS 0.07116 m、旋转 RMS 0.98209°。
+
+这些文件记录的是 host 时间关联版本的现有最终结果。后续采用 RTK 测量时间重新导出后，应写入新的结果目录，不能覆盖本目录。
@@ -1,45 +0,0 @@
-{
-  "convention": "T_RTK_lidar maps raw LiDAR points into the RTK navigation frame: p_RTK = T_RTK_lidar * p_lidar",
-  "translation_m": [
-    1.6381793500373911,
-    -0.24084479868828831,
-    0.08448123595331278
-  ],
-  "rotation_rpy_deg_xyz": [
-    -0.8171674587248069,
-    1.323288118779805,
-    -22.104163317857477
-  ],
-  "quaternion_xyzw": [
-    -0.004784711987091957,
-    0.012700096588977744,
-    -0.19160273666049008,
-    0.9813787267829084
-  ],
-  "matrix_4x4": [
-    [
-      0.926254197701683,
-      0.37594816689521215,
-      0.026760737062740007,
-      1.6381793500373911
-    ],
-    [
-      -0.3761912321127582,
-      0.926530995670823,
-      0.004524482591229066,
-      -0.24084479868828831
-    ],
-    [
-      -0.02309368141930373,
-      -0.014257975640431828,
-      0.9996316281556624,
-      0.08448123595331278
-    ],
-    [
-      0.0,
-      0.0,
-      0.0,
-      1.0
-    ]
-  ]
-}
@@ -1,23 +0,0 @@
-# data4参考结果
-
-推荐下游只读取`final_T_RTK_lidar.json`，其方向为：
-
-```text
-p_RTK = T_RTK_lidar · p_lidar
-```
-
-```text
-translation_m = [1.638179350, -0.240844799, 0.084481236]
-RPY_deg_xyz   = [-0.817167459, 1.323288119, -22.104163318]
-```
-
-| 路径 | 内容 |
-|---|---|
-| `final_T_RTK_lidar.json` | 唯一推荐使用的最终外参 |
-| `summary.json` | 最终残差、条件数和两后端差异摘要 |
-| `common/ground_planes.csv` | 34站地面RANSAC平面 |
-| `open3d_gicp/` | Open3D精筛B、精筛审计、逐对质量CSV和独立X |
-| `small_gicp/` | small_gicp对应产物 |
-| `consensus/` | 两后端共同认可的25对B、共识审计和最终X原始求解记录 |
-
-`z=0.084481 m`依赖RTK参考点离地`0.8535 m`，不是平面AX=XB独立观测值。当前AX RMS约`0.100207 m / 1.252794°`，结果适合算法联调和继续验证，不应据此单独宣称逐帧GT达到±3 cm。
@@ -1,35 +0,0 @@
-time,nx,ny,nz,d,inliers,rms_m,frame_counter
-1784783825.357129,-0.0071009788712051635,-0.01614775434066578,0.9998444009588814,0.9449714797885561,2216,0.01353681235386204,382
-1784783905.353819,0.0037577953662433442,-0.00645689216229629,0.999972093369405,0.9404093678203617,1992,0.01182484680356077,1182
-1784783971.0503054,-0.021571557237587136,-0.004187980739348835,0.9997585352152151,0.9464426916222599,1922,0.01241302113439391,1839
-1784784059.7468228,-0.02328381206588687,0.004386003115594592,0.999719274132669,0.9506398743758987,1825,0.013207042662497559,2726
-1784784149.2434597,-0.03034651356298797,8.407602826756324e-05,0.9995394349628197,0.9287260086761917,1631,0.012431422856007433,3621
-1784784224.2408776,-0.02735108660242708,0.010336463557273514,0.9995724463903534,0.8697042041945898,1745,0.012010699567274814,4371
-1784784301.6372502,-0.008085199791897242,-0.01261499998640764,0.9998877393586082,0.9555128377061561,2057,0.011385954851802133,5145
-1784784387.733771,-0.007167201772462397,-0.008280272080485561,0.9999400323584541,0.9272403036781977,2190,0.012588169113362788,6006
-1784784474.9314597,-0.015033673776128801,-0.05115510005163043,0.9985775605287256,0.9072885818985176,1852,0.011643717831315507,6878
-1784784549.4274275,-0.03189811408163763,0.004490776571011993,0.999481036960594,0.9463330979345341,1638,0.013973864979699106,7623
-1784784614.7244046,-0.027669126419476022,-0.013618714393779237,0.999524361914928,0.9348654978703125,1794,0.011102769699524444,8276
-1784784682.921899,-0.022151829941254066,-0.026156017173182243,0.9994124069651578,0.9227652769218206,1638,0.01266204532799032,8958
-1784784758.8187964,-0.022601789108727538,-0.030665670278286643,0.999274124450077,0.9305425654631599,1795,0.013340493954266352,9717
-1784784836.0155501,-0.017598498968453644,-0.02647415434199472,0.999494578267403,0.961769336191532,2015,0.01382548272951725,10489
-1784784921.1126208,-0.021874728045639568,-0.019005922983034846,0.9995800474021539,0.9186945373736978,1808,0.01319739563436461,11340
-1784784992.709947,-0.019622211270580107,-0.02528822634559132,0.9994876059427386,0.9414876680920201,2158,0.013280256398986076,12056
-1784785067.6067727,-0.031060743296391496,-0.010214597374617485,0.9994653031628212,0.9402571806911639,1758,0.013392641298608525,12805
-1784785215.9006598,-0.023673459396853343,-0.027330465291762113,0.9993460926961797,0.9143984233881569,2134,0.012198902426752438,14288
-1784785296.4990919,-0.013579953767407775,-0.025158411654770292,0.9995912360453568,0.929929365058787,2445,0.012790980094197171,15094
-1784785363.1952267,-0.0316919344947604,-0.008177115456525313,0.9994642345130668,0.9579125765731408,1861,0.012743464679231025,15761
-1784785434.592462,-0.022678088603046032,0.004435617789851151,0.9997329791459992,0.9537383917106543,1853,0.013464094518301148,16475
-1784785506.389296,-0.0273694753756875,-0.01771941378886425,0.9994683257575693,0.9327361226688458,1749,0.011502338837958854,17193
-1784785587.5863533,-0.034617559115875766,0.0015843237885758451,0.999399376885433,0.9191959537091571,1744,0.013401267879037225,18005
-1784785681.9825997,-0.033298152875437845,-0.018946770164947627,0.9992658569747096,0.9446967789786688,2037,0.012804059875312601,18949
-1784785815.4779446,-0.006681441650905292,-0.023720354219030532,0.9996963054514052,0.964084392350492,2080,0.013159652224503205,20284
-1784785891.9768085,-0.026044255070688936,0.004021590005713901,0.9996527014876911,0.934003424141447,1610,0.012619787010493816,21049
-1784785967.8726046,-0.026983516555153,-0.01791249491380091,0.9994753785663161,0.9339463871372334,1482,0.013831424226303278,21808
-1784786031.5701303,-0.028810092762610772,-0.015551490666087386,0.9994639211562729,0.938325903591574,1592,0.013415706854842701,22445
-1784786087.9670725,-0.026424364888569394,-0.014575043111831797,0.9995445568150146,0.9345909622822591,1466,0.013090809334738121,23009
-1784786160.8647907,-0.032665212336793446,0.0597663130328534,0.9976777895340014,1.0611447790248285,1511,0.012251618222434443,23738
-1784786252.6621523,-0.03732336904542465,-0.020702346452411244,0.9990887743211128,0.9478808317179221,1719,0.012520822005912273,24656
-1784786319.6581354,-0.025461816426307887,-0.02602901616210242,0.9993368732424047,0.9466783627035876,1589,0.013468160971926036,25326
-1784786396.7558627,-0.025295174442520576,-0.02343833470627969,0.9994052224278793,0.9710592140122102,1321,0.012895976784738191,26097
-1784786557.4492514,-0.014426534652625146,-0.00926546906583815,0.9998530022862894,0.935013905231254,1251,0.012477706451163199,27704
@@ -1,332 +0,0 @@
-{
-  "selection_is_X_independent": true,
-  "B_source": "Open3D; small_gicp is used only as an agreement gate",
-  "max_translation_m": 0.05,
-  "max_rotation_deg": 0.5,
-  "input_open3d_pairs": 41,
-  "accepted_pairs": 25,
-  "pairs": [
-    {
-      "i": 0,
-      "j": 1,
-      "open3d_small_translation_m": 0.014276780914058016,
-      "open3d_small_rotation_deg": 0.6136066194510507,
-      "accepted": false,
-      "reason": "backend_disagreement"
-    },
-    {
-      "i": 0,
-      "j": 2,
-      "open3d_small_translation_m": 0.019952450418738,
-      "open3d_small_rotation_deg": 0.14609270025586996,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 1,
-      "j": 2,
-      "accepted": false,
-      "reason": "not_in_small_gicp_refined"
-    },
-    {
-      "i": 2,
-      "j": 3,
-      "open3d_small_translation_m": 0.022407292900448784,
-      "open3d_small_rotation_deg": 0.1674801169908669,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 2,
-      "j": 5,
-      "open3d_small_translation_m": 0.006161707315193906,
-      "open3d_small_rotation_deg": 0.13760639079130288,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 3,
-      "j": 5,
-      "open3d_small_translation_m": 0.039798937040509075,
-      "open3d_small_rotation_deg": 0.21317157512260662,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 3,
-      "j": 6,
-      "open3d_small_translation_m": 0.012411893826145848,
-      "open3d_small_rotation_deg": 0.6409039547122766,
-      "accepted": false,
-      "reason": "backend_disagreement"
-    },
-    {
-      "i": 6,
-      "j": 7,
-      "open3d_small_translation_m": 0.008481323658875535,
-      "open3d_small_rotation_deg": 0.21102463812659789,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 6,
-      "j": 8,
-      "open3d_small_translation_m": 0.0392892670228761,
-      "open3d_small_rotation_deg": 0.3638880951335145,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 7,
-      "j": 8,
-      "open3d_small_translation_m": 0.008121615288997118,
-      "open3d_small_rotation_deg": 0.6269514061788293,
-      "accepted": false,
-      "reason": "backend_disagreement"
-    },
-    {
-      "i": 10,
-      "j": 11,
-      "open3d_small_translation_m": 0.021311366483595485,
-      "open3d_small_rotation_deg": 0.5895827931090589,
-      "accepted": false,
-      "reason": "backend_disagreement"
-    },
-    {
-      "i": 12,
-      "j": 14,
-      "open3d_small_translation_m": 0.02159358315642783,
-      "open3d_small_rotation_deg": 0.2828054398005336,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 12,
-      "j": 15,
-      "open3d_small_translation_m": 0.036801934207227605,
-      "open3d_small_rotation_deg": 0.5608491395108458,
-      "accepted": false,
-      "reason": "backend_disagreement"
-    },
-    {
-      "i": 13,
-      "j": 15,
-      "open3d_small_translation_m": 0.02340982602704092,
-      "open3d_small_rotation_deg": 0.5560478634176542,
-      "accepted": false,
-      "reason": "backend_disagreement"
-    },
-    {
-      "i": 13,
-      "j": 16,
-      "open3d_small_translation_m": 0.027602744462629097,
-      "open3d_small_rotation_deg": 0.2529072067309812,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 15,
-      "j": 16,
-      "open3d_small_translation_m": 0.02090824470779122,
-      "open3d_small_rotation_deg": 0.034004621265071464,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 15,
-      "j": 17,
-      "open3d_small_translation_m": 0.04186146133502706,
-      "open3d_small_rotation_deg": 0.13358624916951445,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 15,
-      "j": 18,
-      "open3d_small_translation_m": 0.01461017711105615,
-      "open3d_small_rotation_deg": 0.2537851025085552,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 16,
-      "j": 19,
-      "open3d_small_translation_m": 0.012703728170379172,
-      "open3d_small_rotation_deg": 0.36607447371235324,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 17,
-      "j": 18,
-      "open3d_small_translation_m": 0.030547382773647488,
-      "open3d_small_rotation_deg": 0.6293030402470645,
-      "accepted": false,
-      "reason": "backend_disagreement"
-    },
-    {
-      "i": 21,
-      "j": 22,
-      "open3d_small_translation_m": 0.008438605953756149,
-      "open3d_small_rotation_deg": 0.15351222882817242,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 21,
-      "j": 23,
-      "open3d_small_translation_m": 0.0184790436482599,
-      "open3d_small_rotation_deg": 0.5781132894399331,
-      "accepted": false,
-      "reason": "backend_disagreement"
-    },
-    {
-      "i": 21,
-      "j": 24,
-      "open3d_small_translation_m": 0.03128464242771278,
-      "open3d_small_rotation_deg": 0.3921882672789548,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 22,
-      "j": 23,
-      "open3d_small_translation_m": 0.01156826970365892,
-      "open3d_small_rotation_deg": 0.061357836393243825,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 22,
-      "j": 25,
-      "open3d_small_translation_m": 0.04101882834929893,
-      "open3d_small_rotation_deg": 0.420339642399271,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 23,
-      "j": 24,
-      "open3d_small_translation_m": 0.009191025382465435,
-      "open3d_small_rotation_deg": 0.5691342384946491,
-      "accepted": false,
-      "reason": "backend_disagreement"
-    },
-    {
-      "i": 25,
-      "j": 26,
-      "open3d_small_translation_m": 0.02091454005252894,
-      "open3d_small_rotation_deg": 0.4875072590330582,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 25,
-      "j": 27,
-      "open3d_small_translation_m": 0.09632206662156642,
-      "open3d_small_rotation_deg": 0.3699938120965201,
-      "accepted": false,
-      "reason": "backend_disagreement"
-    },
-    {
-      "i": 25,
-      "j": 28,
-      "open3d_small_translation_m": 0.0301381128127144,
-      "open3d_small_rotation_deg": 0.46587172800425497,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 26,
-      "j": 27,
-      "open3d_small_translation_m": 0.01010757207206556,
-      "open3d_small_rotation_deg": 0.4201365669869023,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 26,
-      "j": 28,
-      "open3d_small_translation_m": 0.006243998078352694,
-      "open3d_small_rotation_deg": 1.3246136424341457,
-      "accepted": false,
-      "reason": "backend_disagreement"
-    },
-    {
-      "i": 26,
-      "j": 29,
-      "open3d_small_translation_m": 0.03617825286340023,
-      "open3d_small_rotation_deg": 0.27823666091247784,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 27,
-      "j": 28,
-      "open3d_small_translation_m": 0.008540949499580083,
-      "open3d_small_rotation_deg": 0.3676810414714511,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 27,
-      "j": 29,
-      "open3d_small_translation_m": 0.01399120240175791,
-      "open3d_small_rotation_deg": 0.437290577300882,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 28,
-      "j": 29,
-      "open3d_small_translation_m": 0.12711262123542025,
-      "open3d_small_rotation_deg": 0.9231252582756669,
-      "accepted": false,
-      "reason": "backend_disagreement"
-    },
-    {
-      "i": 28,
-      "j": 30,
-      "open3d_small_translation_m": 0.030113115024945805,
-      "open3d_small_rotation_deg": 0.33608547109429165,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 28,
-      "j": 31,
-      "accepted": false,
-      "reason": "not_in_small_gicp_refined"
-    },
-    {
-      "i": 30,
-      "j": 31,
-      "open3d_small_translation_m": 0.056474627862098614,
-      "open3d_small_rotation_deg": 0.6666837905088073,
-      "accepted": false,
-      "reason": "backend_disagreement"
-    },
-    {
-      "i": 30,
-      "j": 32,
-      "open3d_small_translation_m": 0.0152913302124184,
-      "open3d_small_rotation_deg": 0.3092824993876474,
-      "accepted": true,
-      "reason": ""
-    },
-    {
-      "i": 31,
-      "j": 32,
-      "accepted": false,
-      "reason": "not_in_small_gicp_refined"
-    },
-    {
-      "i": 31,
-      "j": 33,
-      "open3d_small_translation_m": 0.013286737983556427,
-      "open3d_small_rotation_deg": 0.37286489690414826,
-      "accepted": true,
-      "reason": ""
-    }
-  ]
-}
@@ -1,266 +0,0 @@
-{
-  "schema_version": 1,
-  "success": true,
-  "convention": "T_RTK_lidar maps raw LiDAR points into the RTK navigation frame",
-  "equation": "A_RTK_ij X = X B_LiDAR_ij",
-  "frames": {
-    "RTK": {
-      "origin": "GGA positioning reference point; confirm ANT1/reference antenna in receiver configuration",
-      "x_axis": "horizontal projection of the rawHeading baseline direction reported by the receiver",
-      "y_axis": "left",
-      "z_axis": "up",
-      "yaw_enu_deg": "90 - rawHeadingDeg"
-    },
-    "LiDAR": "raw LiDAR sensor frame"
-  },
-  "backend": "consensus",
-  "measured_lidar_extrinsic_used_as_initial": false,
-  "body_heading_offset_used": false,
-  "body_antenna_lever_xy_used": false,
-  "translation_m": [
-    1.6381793500373911,
-    -0.24084479868828831,
-    0.08448123595331278
-  ],
-  "rotation_rpy_deg_xyz": [
-    -0.8171674587248069,
-    1.323288118779805,
-    -22.104163317857477
-  ],
-  "quaternion_xyzw": [
-    -0.004784711987091957,
-    0.012700096588977744,
-    -0.19160273666049008,
-    0.9813787267829084
-  ],
-  "matrix_4x4": [
-    [
-      0.926254197701683,
-      0.37594816689521215,
-      0.026760737062740007,
-      1.6381793500373911
-    ],
-    [
-      -0.3761912321127582,
-      0.926530995670823,
-      0.004524482591229066,
-      -0.24084479868828831
-    ],
-    [
-      -0.02309368141930373,
-      -0.014257975640431828,
-      0.9996316281556624,
-      0.08448123595331278
-    ],
-    [
-      0.0,
-      0.0,
-      0.0,
-      1.0
-    ]
-  ],
-  "quality": {
-    "stations": 34,
-    "pairs": 25,
-    "residuals": {
-      "pairs": 25,
-      "translation_m": {
-        "rms": 0.10020667268070801,
-        "median": 0.06249704966098745,
-        "p90": 0.1210855900297508,
-        "p95": 0.12306740757559503,
-        "max": 0.35253378022021187
-      },
-      "rotation_deg": {
-        "rms": 1.2527941187072538,
-        "median": 0.7469173100535645,
-        "p90": 1.8215127226094046,
-        "p95": 1.973738680622839,
-        "max": 4.325671952515919
-      },
-      "per_pair": [
-        {
-          "pair_index": 0,
-          "translation_m": 0.04167906865173614,
-          "rotation_deg": 0.7044680397451787
-        },
-        {
-          "pair_index": 1,
-          "translation_m": 0.05434432733915097,
-          "rotation_deg": 0.3949090486960162
-        },
-        {
-          "pair_index": 2,
-          "translation_m": 0.10468985225969502,
-          "rotation_deg": 0.2966790038612149
-        },
-        {
-          "pair_index": 3,
-          "translation_m": 0.064702341381345,
-          "rotation_deg": 0.5538785486937366
-        },
-        {
-          "pair_index": 4,
-          "translation_m": 0.06403624975051968,
-          "rotation_deg": 0.7469173100535645
-        },
-        {
-          "pair_index": 5,
-          "translation_m": 0.019589736031301614,
-          "rotation_deg": 0.7060778312055135
-        },
-        {
-          "pair_index": 6,
-          "translation_m": 0.04563086081489213,
-          "rotation_deg": 0.8701331718844921
-        },
-        {
-          "pair_index": 7,
-          "translation_m": 0.05643960299219297,
-          "rotation_deg": 1.8737737246943913
-        },
-        {
-          "pair_index": 8,
-          "translation_m": 0.1234266519889254,
-          "rotation_deg": 1.9987299196049513
-        },
-        {
-          "pair_index": 9,
-          "translation_m": 0.06753558655646176,
-          "rotation_deg": 0.2674085857580904
-        },
-        {
-          "pair_index": 10,
-          "translation_m": 0.057937366822662255,
-          "rotation_deg": 0.339735283375759
-        },
-        {
-          "pair_index": 11,
-          "translation_m": 0.03457457850935045,
-          "rotation_deg": 0.5158041285368985
-        },
-        {
-          "pair_index": 12,
-          "translation_m": 0.06294442058649205,
-          "rotation_deg": 0.6210132605220049
-        },
-        {
-          "pair_index": 13,
-          "translation_m": 0.04800822576662163,
-          "rotation_deg": 1.7431212194819241
-        },
-        {
-          "pair_index": 14,
-          "translation_m": 0.06249704966098745,
-          "rotation_deg": 0.8651715760380532
-        },
-        {
-          "pair_index": 15,
-          "translation_m": 0.12026833019096655,
-          "rotation_deg": 4.325671952515919
-        },
-        {
-          "pair_index": 16,
-          "translation_m": 0.04916024916487916,
-          "rotation_deg": 0.9331159567430728
-        },
-        {
-          "pair_index": 17,
-          "translation_m": 0.02802786916783814,
-          "rotation_deg": 0.8755591753060336
-        },
-        {
-          "pair_index": 18,
-          "translation_m": 0.054001671602238746,
-          "rotation_deg": 0.48777111889991587
-        },
-        {
-          "pair_index": 19,
-          "translation_m": 0.10122861599246656,
-          "rotation_deg": 0.9367994244405716
-        },
-        {
-          "pair_index": 20,
-          "translation_m": 0.023164324685695653,
-          "rotation_deg": 0.2827650802657415
-        },
-        {
-          "pair_index": 21,
-          "translation_m": 0.06461070522327692,
-          "rotation_deg": 1.1522606437088527
-        },
-        {
-          "pair_index": 22,
-          "translation_m": 0.35253378022021187,
-          "rotation_deg": 0.7805430662091233
-        },
-        {
-          "pair_index": 23,
-          "translation_m": 0.11095170748867093,
-          "rotation_deg": 0.8561868088481429
-        },
-        {
-          "pair_index": 24,
-          "translation_m": 0.12163042992227363,
-          "rotation_deg": 0.17586872890763125
-        }
-      ]
-    },
-    "weighted_jacobian_condition_number": 7.739413195936781,
-    "linearized_one_sigma": {
-      "translation_m": [
-        0.008936804232414386,
-        0.009199590403303341,
-        0.004827611183429192
-      ],
-      "rotation_deg": [
-        0.08042474849565617,
-        0.07825800357808266,
-        0.1694976028248396
-      ],
-      "warning": "conditional local estimate; bootstrap is the primary stability check"
-    },
-    "bootstrap": {
-      "runs": 200,
-      "order": [
-        "x_m",
-        "y_m",
-        "z_m",
-        "roll_deg",
-        "pitch_deg",
-        "yaw_deg"
-      ],
-      "std": [
-        0.004007472116212317,
-        0.00447369016783699,
-        0.003161844455893403,
-        0.10272261218646783,
-        0.09618742156951016,
-        0.14509879375374413
-      ],
-      "p025": [
-        1.6309949623104336,
-        -0.2470615833911814,
-        0.07875765037498306,
-        -0.9881676010783537,
-        1.1135179371917805,
-        -22.3851229846878
-      ],
-      "p975": [
-        1.6456999445482652,
-        -0.22784916633142768,
-        0.09114084095723367,
-        -0.5808877804701225,
-        1.4821798187931374,
-        -21.82296828752347
-      ]
-    }
-  },
-  "z_constraint": {
-    "observable_from_planar_AX_XB": false,
-    "method": "LiDAR ground planes plus externally supplied RTK reference-point height above ground",
-    "rtk_reference_height_above_ground_m": 0.8535,
-    "warning": "z is conditional on the supplied RTK antenna height; it is not independently identified by planar Ackermann motion"
-  },
-  "important_limit": "AX residual and bootstrap quantify internal consistency, not independent centimetre-grade absolute certification"
-}
@@ -1,300 +0,0 @@
-{
-  "schema_version": 1,
-  "success": true,
-  "convention": "T_RTK_lidar maps raw LiDAR points into the RTK navigation frame",
-  "equation": "A_RTK_ij X = X B_LiDAR_ij",
-  "frames": {
-    "RTK": {
-      "origin": "GGA positioning reference point; confirm ANT1/reference antenna in receiver configuration",
-      "x_axis": "horizontal projection of the rawHeading baseline direction reported by the receiver",
-      "y_axis": "left",
-      "z_axis": "up",
-      "yaw_enu_deg": "90 - rawHeadingDeg"
-    },
-    "LiDAR": "raw LiDAR sensor frame"
-  },
-  "backend": "consensus",
-  "measured_lidar_extrinsic_used_as_initial": false,
-  "body_heading_offset_used": false,
-  "body_antenna_lever_xy_used": false,
-  "translation_m": [
-    1.6381793500373911,
-    -0.24084479868828831,
-    0.08448123595331278
-  ],
-  "rotation_rpy_deg_xyz": [
-    -0.8171674587248069,
-    1.323288118779805,
-    -22.104163317857477
-  ],
-  "quaternion_xyzw": [
-    -0.004784711987091957,
-    0.012700096588977744,
-    -0.19160273666049008,
-    0.9813787267829084
-  ],
-  "matrix_4x4": [
-    [
-      0.926254197701683,
-      0.37594816689521215,
-      0.026760737062740007,
-      1.6381793500373911
-    ],
-    [
-      -0.3761912321127582,
-      0.926530995670823,
-      0.004524482591229066,
-      -0.24084479868828831
-    ],
-    [
-      -0.02309368141930373,
-      -0.014257975640431828,
-      0.9996316281556624,
-      0.08448123595331278
-    ],
-    [
-      0.0,
-      0.0,
-      0.0,
-      1.0
-    ]
-  ],
-  "quality": {
-    "stations": 34,
-    "pairs": 25,
-    "residuals": {
-      "pairs": 25,
-      "translation_m": {
-        "rms": 0.10020667268070801,
-        "median": 0.06249704966098745,
-        "p90": 0.1210855900297508,
-        "p95": 0.12306740757559503,
-        "max": 0.35253378022021187
-      },
-      "rotation_deg": {
-        "rms": 1.2527941187072538,
-        "median": 0.7469173100535645,
-        "p90": 1.8215127226094046,
-        "p95": 1.973738680622839,
-        "max": 4.325671952515919
-      },
-      "per_pair": [
-        {
-          "pair_index": 0,
-          "translation_m": 0.04167906865173614,
-          "rotation_deg": 0.7044680397451787
-        },
-        {
-          "pair_index": 1,
-          "translation_m": 0.05434432733915097,
-          "rotation_deg": 0.3949090486960162
-        },
-        {
-          "pair_index": 2,
-          "translation_m": 0.10468985225969502,
-          "rotation_deg": 0.2966790038612149
-        },
-        {
-          "pair_index": 3,
-          "translation_m": 0.064702341381345,
-          "rotation_deg": 0.5538785486937366
-        },
-        {
-          "pair_index": 4,
-          "translation_m": 0.06403624975051968,
-          "rotation_deg": 0.7469173100535645
-        },
-        {
-          "pair_index": 5,
-          "translation_m": 0.019589736031301614,
-          "rotation_deg": 0.7060778312055135
-        },
-        {
-          "pair_index": 6,
-          "translation_m": 0.04563086081489213,
-          "rotation_deg": 0.8701331718844921
-        },
-        {
-          "pair_index": 7,
-          "translation_m": 0.05643960299219297,
-          "rotation_deg": 1.8737737246943913
-        },
-        {
-          "pair_index": 8,
-          "translation_m": 0.1234266519889254,
-          "rotation_deg": 1.9987299196049513
-        },
-        {
-          "pair_index": 9,
-          "translation_m": 0.06753558655646176,
-          "rotation_deg": 0.2674085857580904
-        },
-        {
-          "pair_index": 10,
-          "translation_m": 0.057937366822662255,
-          "rotation_deg": 0.339735283375759
-        },
-        {
-          "pair_index": 11,
-          "translation_m": 0.03457457850935045,
-          "rotation_deg": 0.5158041285368985
-        },
-        {
-          "pair_index": 12,
-          "translation_m": 0.06294442058649205,
-          "rotation_deg": 0.6210132605220049
-        },
-        {
-          "pair_index": 13,
-          "translation_m": 0.04800822576662163,
-          "rotation_deg": 1.7431212194819241
-        },
-        {
-          "pair_index": 14,
-          "translation_m": 0.06249704966098745,
-          "rotation_deg": 0.8651715760380532
-        },
-        {
-          "pair_index": 15,
-          "translation_m": 0.12026833019096655,
-          "rotation_deg": 4.325671952515919
-        },
-        {
-          "pair_index": 16,
-          "translation_m": 0.04916024916487916,
-          "rotation_deg": 0.9331159567430728
-        },
-        {
-          "pair_index": 17,
-          "translation_m": 0.02802786916783814,
-          "rotation_deg": 0.8755591753060336
-        },
-        {
-          "pair_index": 18,
-          "translation_m": 0.054001671602238746,
-          "rotation_deg": 0.48777111889991587
-        },
-        {
-          "pair_index": 19,
-          "translation_m": 0.10122861599246656,
-          "rotation_deg": 0.9367994244405716
-        },
-        {
-          "pair_index": 20,
-          "translation_m": 0.023164324685695653,
-          "rotation_deg": 0.2827650802657415
-        },
-        {
-          "pair_index": 21,
-          "translation_m": 0.06461070522327692,
-          "rotation_deg": 1.1522606437088527
-        },
-        {
-          "pair_index": 22,
-          "translation_m": 0.35253378022021187,
-          "rotation_deg": 0.7805430662091233
-        },
-        {
-          "pair_index": 23,
-          "translation_m": 0.11095170748867093,
-          "rotation_deg": 0.8561868088481429
-        },
-        {
-          "pair_index": 24,
-          "translation_m": 0.12163042992227363,
-          "rotation_deg": 0.17586872890763125
-        }
-      ]
-    },
-    "weighted_jacobian_condition_number": 7.739413195936781,
-    "linearized_one_sigma": {
-      "translation_m": [
-        0.008936804232414386,
-        0.009199590403303341,
-        0.004827611183429192
-      ],
-      "rotation_deg": [
-        0.08042474849565617,
-        0.07825800357808266,
-        0.1694976028248396
-      ],
-      "warning": "conditional local estimate; bootstrap is the primary stability check"
-    },
-    "bootstrap": {
-      "runs": 200,
-      "order": [
-        "x_m",
-        "y_m",
-        "z_m",
-        "roll_deg",
-        "pitch_deg",
-        "yaw_deg"
-      ],
-      "std": [
-        0.004007472116212317,
-        0.00447369016783699,
-        0.003161844455893403,
-        0.10272261218646783,
-        0.09618742156951016,
-        0.14509879375374413
-      ],
-      "p025": [
-        1.6309949623104336,
-        -0.2470615833911814,
-        0.07875765037498306,
-        -0.9881676010783537,
-        1.1135179371917805,
-        -22.3851229846878
-      ],
-      "p975": [
-        1.6456999445482652,
-        -0.22784916633142768,
-        0.09114084095723367,
-        -0.5808877804701225,
-        1.4821798187931374,
-        -21.82296828752347
-      ]
-    }
-  },
-  "z_constraint": {
-    "observable_from_planar_AX_XB": false,
-    "method": "LiDAR ground planes plus externally supplied RTK reference-point height above ground",
-    "rtk_reference_height_above_ground_m": 0.8535,
-    "warning": "z is conditional on the supplied RTK antenna height; it is not independently identified by planar Ackermann motion"
-  },
-  "important_limit": "AX residual and bootstrap quantify internal consistency, not independent centimetre-grade absolute certification",
-  "selection": {
-    "recommended": true,
-    "reason": "Uses only motion pairs accepted independently by both Open3D GICP and small_gicp",
-    "open3d_vs_small_gicp": {
-      "translation_m": 0.003889255293759414,
-      "rotation_deg": 0.1884307130161592,
-      "delta_matrix_4x4": [
-        [
-          0.9999968771376496,
-          0.0024968749848742764,
-          -0.00010644368722136346,
-          0.0008526003522697501
-        ],
-        [
-          -0.0024970968314049877,
-          0.9999945974960792,
-          -0.0021376356258303525,
-          -0.003658904827736509
-        ],
-        [
-          0.00010110570323801577,
-          0.0021378947504826257,
-          0.9999977095892133,
-          0.0010058801324768218
-        ],
-        [
-          0.0,
-          0.0,
-          0.0,
-          1.0
-        ]
-      ]
-    }
-  }
-}
@@ -1,97 +0,0 @@
-i,j,rtk_translation_m,rtk_rotation_deg,heldout_inlier_ratio,heldout_inlier_rmse_m,hessian_rank,hessian_condition,reverse_translation_m,reverse_rotation_deg,multistart_success_rate,accepted,rejection_reasons
-0,1,1.6721594124489447,24.171297449440814,0.8061657032755298,0.10961296014103396,6,2.7038608113687213,0.004225163540003575,0.1530449067720668,1.0,True,
-0,2,2.0412175279332088,80.09074797031303,0.7489394523717702,0.116305716193008,6,3.0720058333957327,0.02073003109723684,0.12418344306819311,1.0,True,
-0,3,6.30529961936688,79.9158388329924,0.6310283235519265,0.12470979645173097,6,5.235632990817998,0.02154775870989652,0.25399044979598373,1.0,True,
-1,2,1.2843386040405174,55.9194505208722,0.7867383512544803,0.11079021393934946,6,3.282529873989144,0.00768232673944143,0.0508043300468843,1.0,True,
-1,3,5.433888607887495,55.744541383551606,0.6794562317367552,0.11907169138096609,6,4.183386002322132,0.024832409186708038,0.29364088503444125,1.0,True,
-1,4,1.5299424710613851,106.08652205569952,0.6786112833230006,0.1125501582315264,6,3.2941312581877567,0.0077657602443488094,0.07274574571529673,1.0,True,
-2,3,4.340252832276203,0.1749091373206093,0.7586776859504132,0.11541610277862546,6,3.730803369806122,0.007879904085790266,0.11659483159927261,1.0,True,
-2,4,0.2520257253555564,50.1670715348273,0.7854572527608884,0.1098649389241602,6,2.641287751481567,0.017042953274276868,0.09383247792992644,1.0,True,
-2,5,5.8286926576588955,8.735318060700322,0.7074574574574575,0.11912871456224486,6,3.8527093190832513,0.016640250279170064,0.24773491621516538,1.0,True,
-3,4,4.1070657333447205,50.34198067214791,0.6974624291697462,0.11727630888949149,6,3.5010457716923216,0.0121487212194689,0.20793086843827258,1.0,True,
-3,5,2.2886212019715484,8.910227198020936,0.7962985964476462,0.10646082199215787,6,3.037745853837991,0.011008298723512349,0.0821707761041294,1.0,True,
-3,6,2.618668779147775,47.63555775102663,0.8376509054325956,0.10956583416752531,6,3.1930663579156175,0.001222821038315667,0.092075215117626,1.0,True,
-4,5,5.5767898078953735,41.431753474126985,0.6652516676773802,0.12322094568315027,6,4.985227704290953,0.005399786589871835,0.13893507775898076,0.0,False,multistart_instability
-4,6,6.68811709459501,97.97753842317455,0.04910385465259023,0.15664415071522125,6,4.835889397197473,2.574198069897065,12.647326063758534,0.0,False,heldout_inlier_ratio;forward_reverse_translation;forward_reverse_rotation;multistart_instability
-4,7,6.153247042777442,123.28910472998353,0.020756115641215715,0.16997351387143192,6,17.054474046975617,5.52282870653845,6.186988956437115,0.0,False,heldout_inlier_ratio;heldout_inlier_rmse;forward_reverse_translation;forward_reverse_rotation;multistart_instability
-5,6,2.0562758992403367,56.545784949047565,0.7526921648718901,0.10924852668609375,6,3.289802074469562,0.012374747586500019,0.14461856068328793,1.0,True,
-5,7,0.5812996709001959,81.85735125585653,0.7833561729164071,0.11683079277948678,6,2.8807765869032624,0.013689874812447942,0.20006562776472953,1.0,True,
-5,8,2.6887856969568644,172.47951556359513,0.3979730564825114,0.13180114830799514,6,4.4867592164383865,2.902311115345869,2.1073169352638135,1.0,False,forward_reverse_translation;forward_reverse_rotation
-6,7,1.9723544820714844,25.311566306808967,0.8497729566094854,0.10415909908074775,6,3.3759361297847534,0.008986805974950147,0.13099479506412062,1.0,True,
-6,8,0.7288530799256238,115.93373061454484,0.7678928928928929,0.1116672507978127,6,3.260920573387359,0.010085391730567652,0.1175283699615622,1.0,True,
-6,9,7.898758206937296,103.90638727582184,0.6071384156199477,0.12594282886521943,6,6.882498483502542,2.7065494720876333,9.167174886516003,1.0,False,forward_reverse_translation;forward_reverse_rotation
-7,8,2.6747845281541447,90.62216430773587,0.8299748110831234,0.10748525688830211,6,3.0071179226782405,0.00710646197885058,0.1069872847368705,1.0,True,
-7,9,8.06955581661561,78.59482096901287,0.6188509200150206,0.12713205078393502,6,6.716979637303372,6.104542218165768,6.596313811797356,0.0,False,forward_reverse_translation;forward_reverse_rotation;multistart_instability
-7,10,8.088675434881791,134.6003195530505,0.04729478766868887,0.16509079956796627,6,8.903210909129099,5.698690438550839,24.43813017288333,0.0,False,heldout_inlier_ratio;heldout_inlier_rmse;forward_reverse_translation;forward_reverse_rotation;multistart_instability
-8,9,7.73033999229505,12.027343338722995,0.6326834719980131,0.12084265385763364,6,5.53968988049318,2.4349995045990127,17.158034002952295,1.0,False,forward_reverse_translation;forward_reverse_rotation
-8,10,8.378411682322204,43.97815524531458,0.6163861933423412,0.1293602846396598,6,4.789005902863083,0.011327540721531722,0.17059480981566605,0.0,False,multistart_instability
-8,11,11.214115106391473,22.60670813095403,0.035782503501846426,0.17440177580299876,6,9.744579276034784,1.861364410474017,5.099464012971126,1.0,False,heldout_inlier_ratio;heldout_inlier_rmse;forward_reverse_translation;forward_reverse_rotation
-9,10,1.9264207261313253,56.00549858403757,0.6543345543345543,0.10729272360686744,6,3.3788563349731624,1.6360232143180424,5.55466982065639,1.0,False,forward_reverse_translation;forward_reverse_rotation
-9,11,4.747620352849464,34.63405146967702,0.5818780055682106,0.1171768781510077,6,3.555986532172078,0.00981302583568424,0.06035460198954135,1.0,True,
-9,12,7.208566899019793,16.356227217122623,0.5379123584441162,0.12645695585785705,6,4.882558096197047,0.008862930161051686,0.11566409877698092,1.0,True,
-10,11,3.0731501091601263,21.371447114360553,0.7118898623279099,0.1192865608074921,6,3.220203471557625,0.0029529340190141227,0.004176908093440082,1.0,True,
-10,12,6.011368280631378,39.649271366914945,0.6120311738918656,0.12525652128126136,6,4.691686416856208,0.005686910809247203,0.10204044727299268,1.0,True,
-10,13,9.76425648692991,72.41150002956132,0.516551290119572,0.1354903305714048,6,7.346543684414896,0.013565042721589003,0.32552304340992394,1.0,True,
-11,12,3.3258193587172027,18.277824252554396,0.650555275113579,0.12104882943540898,6,4.612247786063477,0.003583199374507776,0.03300207707174117,1.0,True,
-11,13,7.195203402213438,51.04005291520078,0.5698054068172914,0.1285866274322162,6,7.539783468898048,0.016001435752898144,0.11059625579950419,1.0,True,
-11,14,3.634158560526847,20.548768889074672,0.6420881321982974,0.12414062948335593,6,4.952265047266808,0.012369101516230316,0.01870506040605898,1.0,True,
-12,13,3.8697507070400543,32.762228662646386,0.6972966112450819,0.1168089621311632,6,4.28224799374136,0.01141023876783206,0.04779713993428384,1.0,True,
-12,14,0.9871080784265185,2.2709446365202766,0.8749086479902558,0.09521299965540625,6,3.309695139564422,0.006159472215773367,0.014929948455569534,1.0,True,
-12,15,4.171948395051693,25.86371030092923,0.7022030893897189,0.11797995277580106,6,4.277232786772136,0.008495008365046321,0.10278299144703042,1.0,True,
-13,14,3.7992314627329202,30.491284026126113,0.6955810147299509,0.11455956122705321,6,3.350289886810734,0.01022866463344607,0.03515966054944392,1.0,True,
-13,15,0.9105848166450461,6.898518361717151,0.868300353819945,0.1045421356808481,6,3.2437357133954023,0.002375025382773949,0.01072504764419793,1.0,True,
-13,16,3.4957081323467,18.94489979461447,0.7265456392027422,0.1096252966406241,6,3.5227514244456217,0.008958927594996346,0.03041438512426757,1.0,True,
-14,15,3.8816793634199405,23.592765664408958,0.7120070334086913,0.11868441290330703,6,4.62059246950246,0.002571958018980028,0.055069197511519646,1.0,True,
-14,16,7.29101265002769,49.43618382074057,0.5918615984405458,0.12229328437386515,6,7.149509813179227,0.014273957859025563,0.25325650727957555,1.0,True,
-14,17,5.915950814087913,0.8461207481731609,0.6694009445687298,0.12443900216431925,6,5.1577414296960065,0.0178952010004604,0.10920228290609924,1.0,True,
-15,16,3.579287497246505,25.843418156331627,0.702887537993921,0.11495230769293859,6,3.5402899763525375,0.013545291843396808,0.03346625178333291,1.0,True,
-15,17,2.2400625117908257,24.438886412582114,0.7429531936901991,0.11679524427533863,6,3.5266643942801554,0.009894110027911674,0.07786907370565768,1.0,True,
-15,18,4.6956742726068,3.452521908779405,0.7209645010046886,0.11716134583909138,6,4.125231895423439,0.01165472931184747,0.13683586564190106,1.0,True,
-16,17,2.956793995513645,50.28230456891372,0.618922305764411,0.11254196340940027,6,4.068632188828398,0.031047860213503222,0.10375098145236057,1.0,True,
-16,18,3.369822545690391,22.390896247552213,0.6890156918687589,0.11084024896736888,6,4.421167108842175,0.01556225520373068,0.02495881796885156,1.0,True,
-16,19,2.313038703742191,30.035090485266096,0.8685060899826,0.10135543575024479,6,2.9200722330018793,0.0029522513838272538,0.02753369989558306,1.0,True,
-17,18,2.517968959880004,27.89140832136152,0.7697708305735859,0.10648049893472189,6,3.792822356453168,0.010582276181446961,0.03959905194910384,1.0,True,
-17,19,3.518310045065406,80.31739505417983,0.6293759512937596,0.10954497717502036,6,3.648306929393189,0.012240937390578075,0.06030618467886912,1.0,True,
-17,20,3.4199679241812992,152.98392843416642,0.6014520938674964,0.12300562605352787,6,4.719447385686111,0.03231013197809958,0.12856862709639913,0.0,False,multistart_instability
-18,19,2.0416624616211574,52.42598673281832,0.6827314510833881,0.10834806615100022,6,3.5073857685652805,0.012418496053909043,0.11905018881087433,1.0,True,
-18,20,5.836864764777489,125.0925201128049,0.5756313809779688,0.1265759478434111,6,5.389095141151799,0.012917057356044254,0.2404237301134517,0.0,False,multistart_instability
-18,21,10.84206419743439,175.70238585457497,0.2352252017703723,0.15182541744787395,6,9.65894418804011,0.2039286327273425,0.13690658217533308,0.0,False,heldout_inlier_ratio;forward_reverse_translation;multistart_instability
-19,20,6.120105723142265,72.6665333799865,0.6234734541714874,0.1277530580405749,6,5.958603794025071,0.009237066767190974,0.22540488888103255,1.0,True,
-19,21,11.201089261967727,123.27639912174082,0.3788200074840963,0.1468780339612994,6,10.170069582593085,4.607025297377655,2.696881779819613,1.0,False,forward_reverse_translation;forward_reverse_rotation
-19,22,13.962230939038237,128.85294276465584,0.17798277982779828,0.15985830060250797,6,11.806596815871064,2.332292389085354,2.4044448240559984,0.0,False,heldout_inlier_ratio;forward_reverse_translation;forward_reverse_rotation;multistart_instability
-20,21,5.091648376409225,50.60986574175432,0.6596992097884272,0.12147955429086157,6,4.082032735955382,0.014725537493639118,0.19884224722867958,1.0,True,
-20,22,7.842914217245693,56.186409384669375,0.5739414499308958,0.13255415786946775,6,5.3284863413522086,0.006040617548523686,0.19706632354686843,1.0,True,
-20,23,4.953146569484805,70.79178325235414,0.6456945156330087,0.12075756038041646,6,3.9642702950866346,0.02294901101348451,0.19112109479244785,1.0,True,
-21,22,2.836151129858731,5.576543642915048,0.7716237647919971,0.1163227135854156,6,2.7962015303291894,0.009963578798274064,0.1533289219532281,1.0,True,
-21,23,1.4635995041292513,20.181917510599828,0.8576224819696593,0.09865676260631218,6,3.036795069384516,0.00392257332509571,0.00898487649366286,1.0,True,
-21,24,2.7001854883863183,54.59467208208592,0.7715940569126165,0.11361504553552869,6,2.7681582493333527,0.007475673605592584,0.04227769001294981,1.0,True,
-22,23,3.806551883906871,14.605373867684776,0.7396689147762109,0.11702962848624102,6,3.414804725088913,0.018140159434305195,0.1184732181610567,1.0,True,
-22,24,4.999470711928798,49.01812843917085,0.6983240223463687,0.11898873157361631,6,3.633465593686572,1.9214516862996405,12.118772315210817,1.0,False,forward_reverse_translation;forward_reverse_rotation
-22,25,2.281002791409386,12.391903814042275,0.736861094407697,0.1158639471011007,6,2.383007054117406,0.01855955252477569,0.06452602797902846,1.0,True,
-23,24,1.2663665558774873,34.41275457148608,0.7853164556962026,0.11287254423109305,6,2.4099218288471635,0.002195759849349955,0.03211295915781923,1.0,True,
-23,25,5.050865141821003,2.213470053642503,0.6881127450980392,0.12303206700403985,6,2.936372721803798,0.004939188019097873,0.12964064637099942,1.0,True,
-23,26,5.595299803053147,40.730927532824325,0.6852618757612667,0.12040544972155913,6,3.059006509397719,0.006032445250251169,0.14039335223022864,1.0,True,
-24,25,6.316196707646632,36.626224625128586,0.677667493796526,0.1234631580581415,6,3.6286524357748364,0.023479226891170584,0.16273859629355136,1.0,True,
-24,26,6.840027061556236,6.318172961338242,0.6530209617755857,0.12710147612984235,6,3.5331859775372023,0.013311199903813952,0.18193579850150715,1.0,True,
-24,27,7.477875812552711,31.60254826458195,0.6649014778325123,0.12481159614427853,6,4.10006355979974,0.01787282120544869,0.1680556511623414,1.0,True,
-25,26,1.433673687807028,42.944397586466835,0.9127837514934289,0.09581429165893823,6,2.9017658682436958,0.0035341488401767693,0.018440169300164816,1.0,True,
-25,27,1.973107678535245,68.22877288971054,0.8510739856801909,0.10418150333394123,6,2.3682534170899983,0.006103232696003387,0.13366555345654282,1.0,True,
-25,28,2.578633669986193,88.09106909546726,0.8853518429870751,0.10761251229651997,6,2.6372860976794636,0.0034832316659127254,0.03697744201993421,1.0,True,
-26,27,0.6711664435459649,25.284375303243706,0.9183867141162515,0.09074909570650827,6,2.8142315359282533,0.00040199505815399084,0.011554314408123986,1.0,True,
-26,28,1.202247282991071,45.146671509000434,0.8853200095170116,0.1060418493965508,6,2.65640202778952,0.007320240476184076,0.12930335868606002,1.0,True,
-26,29,0.8313560685788559,87.41573662125148,0.7880466815984911,0.10871274240898261,6,3.111957466886598,0.006395243061058376,0.039517035902872893,1.0,True,
-27,28,0.6147316450225401,19.862296205756735,0.9289448669201521,0.08505095515326752,6,2.9632528684698194,0.005157655135593023,0.02266680787731125,1.0,True,
-27,29,0.7540837235696874,62.13136131800778,0.7872365477452019,0.1045551346483567,6,2.9621627623005304,0.011650639607012715,0.16247191340775555,1.0,True,
-27,30,5.07252652550922,152.0939255621477,0.03871268656716418,0.16956979514477974,6,145.85449900829832,4.5477385995336626,37.97511303619642,0.0,False,heldout_inlier_ratio;heldout_inlier_rmse;forward_reverse_translation;forward_reverse_rotation;multistart_instability
-28,29,1.3242039147826599,42.26906511225104,0.8000944621560987,0.10865475473581254,6,3.146622410005858,0.001139416496730335,0.02119627748722763,1.0,True,
-28,30,5.091487440029177,132.23162935639098,0.7721000935453695,0.10815845248211022,6,2.292141242686861,0.004462522533166182,0.03583195432667579,1.0,True,
-28,31,6.538757407908195,158.84212980112872,0.7465330381074466,0.10669666534392452,6,2.40259836682247,0.0063576490595817345,0.04001661521442668,1.0,True,
-29,30,4.767831371266539,89.96256424413991,0.7248812145092132,0.10870831469083345,6,2.947122380473709,0.004656385377159087,0.12465763189228557,0.0,False,multistart_instability
-29,31,5.715598450796842,116.57306468887764,0.7243012243012243,0.11138751941762652,6,2.7645021322697088,0.0042639063218924715,0.04663739608639213,0.0,False,multistart_instability
-29,32,5.281749147864957,159.39493219688646,0.32491640724086246,0.14745892800230104,6,2.8081088899077167,4.8118414680497805,1.8382424276135385,1.0,False,heldout_inlier_ratio;forward_reverse_translation;forward_reverse_rotation
-30,31,2.604154101624297,26.610500444737717,0.8185562292643862,0.09615795732951272,6,2.899558208444641,0.0049991634555749285,0.021384795873435915,1.0,True,
-30,32,1.69105635082049,69.43236795274656,0.8041343079031521,0.09953343153684206,6,2.614616874224757,0.0031713764951448154,0.023767377748422268,1.0,True,
-30,33,3.2411277385703925,160.7145717128097,0.05469213429825602,0.15997514255683,6,6.323554164510002,5.32763716297551,13.705863888332022,0.0,False,heldout_inlier_ratio;forward_reverse_translation;forward_reverse_rotation;multistart_instability
-31,32,0.9137201114724802,42.82186750800884,0.8157085941946499,0.09925944770258255,6,3.048541140076824,0.004719596234885736,0.06146741991683861,1.0,True,
-31,33,1.4965271484681508,134.10407126807198,0.760345235280208,0.09698073659477859,6,2.550150787416455,0.0019366659831763946,0.026538404789073603,1.0,True,
-32,33,1.9169426499676907,91.28220376006315,0.7569928006609229,0.10008740880870787,6,3.6611044282667184,2.9881933505092046,2.6964434945984053,0.0,False,forward_reverse_translation;forward_reverse_rotation;multistart_instability
@@ -1,889 +0,0 @@
-{
-  "selection_is_X_independent": true,
-  "criteria": {
-    "min_inlier_ratio": 0.7,
-    "max_inlier_rmse_m": 0.13,
-    "max_rotation_invariant_error_deg": 0.75,
-    "reverse_translation_tolerance_m": 0.05,
-    "reverse_rotation_tolerance_deg": 0.5
-  },
-  "input_pairs": 73,
-  "accepted_pairs": 41,
-  "pairs": [
-    {
-      "i": 0,
-      "j": 1,
-      "heldout_inlier_ratio": 0.8061657032755298,
-      "heldout_inlier_rmse_m": 0.10961296014103396,
-      "rotation_invariant_error_deg": 0.021953694767642418,
-      "reverse_translation_m": 0.004225163540003575,
-      "reverse_rotation_deg": 0.1530449067720668,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 0,
-      "j": 2,
-      "heldout_inlier_ratio": 0.7489394523717702,
-      "heldout_inlier_rmse_m": 0.116305716193008,
-      "rotation_invariant_error_deg": 0.672251363597141,
-      "reverse_translation_m": 0.02073003109723684,
-      "reverse_rotation_deg": 0.12418344306819311,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 0,
-      "j": 3,
-      "heldout_inlier_ratio": 0.6310283235519265,
-      "heldout_inlier_rmse_m": 0.12470979645173097,
-      "rotation_invariant_error_deg": 0.3195502906704917,
-      "reverse_translation_m": 0.02154775870989652,
-      "reverse_rotation_deg": 0.25399044979598373,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio"
-      ]
-    },
-    {
-      "i": 1,
-      "j": 2,
-      "heldout_inlier_ratio": 0.7867383512544803,
-      "heldout_inlier_rmse_m": 0.11079021393934946,
-      "rotation_invariant_error_deg": 0.6532823847336218,
-      "reverse_translation_m": 0.00768232673944143,
-      "reverse_rotation_deg": 0.0508043300468843,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 1,
-      "j": 3,
-      "heldout_inlier_ratio": 0.6794562317367552,
-      "heldout_inlier_rmse_m": 0.11907169138096609,
-      "rotation_invariant_error_deg": 0.28968149037613955,
-      "reverse_translation_m": 0.024832409186708038,
-      "reverse_rotation_deg": 0.29364088503444125,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio"
-      ]
-    },
-    {
-      "i": 1,
-      "j": 4,
-      "heldout_inlier_ratio": 0.6786112833230006,
-      "heldout_inlier_rmse_m": 0.1125501582315264,
-      "rotation_invariant_error_deg": 0.23509206217087808,
-      "reverse_translation_m": 0.0077657602443488094,
-      "reverse_rotation_deg": 0.07274574571529673,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio"
-      ]
-    },
-    {
-      "i": 2,
-      "j": 3,
-      "heldout_inlier_ratio": 0.7586776859504132,
-      "heldout_inlier_rmse_m": 0.11541610277862546,
-      "rotation_invariant_error_deg": 0.06385020919199427,
-      "reverse_translation_m": 0.007879904085790266,
-      "reverse_rotation_deg": 0.11659483159927261,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 2,
-      "j": 4,
-      "heldout_inlier_ratio": 0.7854572527608884,
-      "heldout_inlier_rmse_m": 0.1098649389241602,
-      "rotation_invariant_error_deg": 0.8714733340459802,
-      "reverse_translation_m": 0.017042953274276868,
-      "reverse_rotation_deg": 0.09383247792992644,
-      "accepted": false,
-      "rejection_reasons": [
-        "rotation_conjugacy_invariant"
-      ]
-    },
-    {
-      "i": 2,
-      "j": 5,
-      "heldout_inlier_ratio": 0.7074574574574575,
-      "heldout_inlier_rmse_m": 0.11912871456224486,
-      "rotation_invariant_error_deg": 0.18237928888532906,
-      "reverse_translation_m": 0.016640250279170064,
-      "reverse_rotation_deg": 0.24773491621516538,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 3,
-      "j": 4,
-      "heldout_inlier_ratio": 0.6974624291697462,
-      "heldout_inlier_rmse_m": 0.11727630888949149,
-      "rotation_invariant_error_deg": 0.4902436735823201,
-      "reverse_translation_m": 0.0121487212194689,
-      "reverse_rotation_deg": 0.20793086843827258,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio"
-      ]
-    },
-    {
-      "i": 3,
-      "j": 5,
-      "heldout_inlier_ratio": 0.7962985964476462,
-      "heldout_inlier_rmse_m": 0.10646082199215787,
-      "rotation_invariant_error_deg": 0.5522737825084345,
-      "reverse_translation_m": 0.011008298723512349,
-      "reverse_rotation_deg": 0.0821707761041294,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 3,
-      "j": 6,
-      "heldout_inlier_ratio": 0.8376509054325956,
-      "heldout_inlier_rmse_m": 0.10956583416752531,
-      "rotation_invariant_error_deg": 0.24026261153986894,
-      "reverse_translation_m": 0.001222821038315667,
-      "reverse_rotation_deg": 0.092075215117626,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 5,
-      "j": 6,
-      "heldout_inlier_ratio": 0.7526921648718901,
-      "heldout_inlier_rmse_m": 0.10924852668609375,
-      "rotation_invariant_error_deg": 0.8049600839827491,
-      "reverse_translation_m": 0.012374747586500019,
-      "reverse_rotation_deg": 0.14461856068328793,
-      "accepted": false,
-      "rejection_reasons": [
-        "rotation_conjugacy_invariant"
-      ]
-    },
-    {
-      "i": 5,
-      "j": 7,
-      "heldout_inlier_ratio": 0.7833561729164071,
-      "heldout_inlier_rmse_m": 0.11683079277948678,
-      "rotation_invariant_error_deg": 0.7937418989476015,
-      "reverse_translation_m": 0.013689874812447942,
-      "reverse_rotation_deg": 0.20006562776472953,
-      "accepted": false,
-      "rejection_reasons": [
-        "rotation_conjugacy_invariant"
-      ]
-    },
-    {
-      "i": 6,
-      "j": 7,
-      "heldout_inlier_ratio": 0.8497729566094854,
-      "heldout_inlier_rmse_m": 0.10415909908074775,
-      "rotation_invariant_error_deg": 0.026617733421641532,
-      "reverse_translation_m": 0.008986805974950147,
-      "reverse_rotation_deg": 0.13099479506412062,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 6,
-      "j": 8,
-      "heldout_inlier_ratio": 0.7678928928928929,
-      "heldout_inlier_rmse_m": 0.1116672507978127,
-      "rotation_invariant_error_deg": 0.32815442351946444,
-      "reverse_translation_m": 0.010085391730567652,
-      "reverse_rotation_deg": 0.1175283699615622,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 7,
-      "j": 8,
-      "heldout_inlier_ratio": 0.8299748110831234,
-      "heldout_inlier_rmse_m": 0.10748525688830211,
-      "rotation_invariant_error_deg": 0.29848794137558343,
-      "reverse_translation_m": 0.00710646197885058,
-      "reverse_rotation_deg": 0.1069872847368705,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 9,
-      "j": 11,
-      "heldout_inlier_ratio": 0.5818780055682106,
-      "heldout_inlier_rmse_m": 0.1171768781510077,
-      "rotation_invariant_error_deg": 0.2635280280577561,
-      "reverse_translation_m": 0.00981302583568424,
-      "reverse_rotation_deg": 0.06035460198954135,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio"
-      ]
-    },
-    {
-      "i": 9,
-      "j": 12,
-      "heldout_inlier_ratio": 0.5379123584441162,
-      "heldout_inlier_rmse_m": 0.12645695585785705,
-      "rotation_invariant_error_deg": 0.5448949372769505,
-      "reverse_translation_m": 0.008862930161051686,
-      "reverse_rotation_deg": 0.11566409877698092,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio"
-      ]
-    },
-    {
-      "i": 10,
-      "j": 11,
-      "heldout_inlier_ratio": 0.7118898623279099,
-      "heldout_inlier_rmse_m": 0.1192865608074921,
-      "rotation_invariant_error_deg": 0.037876295004018345,
-      "reverse_translation_m": 0.0029529340190141227,
-      "reverse_rotation_deg": 0.004176908093440082,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 10,
-      "j": 12,
-      "heldout_inlier_ratio": 0.6120311738918656,
-      "heldout_inlier_rmse_m": 0.12525652128126136,
-      "rotation_invariant_error_deg": 0.22694248875488654,
-      "reverse_translation_m": 0.005686910809247203,
-      "reverse_rotation_deg": 0.10204044727299268,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio"
-      ]
-    },
-    {
-      "i": 10,
-      "j": 13,
-      "heldout_inlier_ratio": 0.516551290119572,
-      "heldout_inlier_rmse_m": 0.1354903305714048,
-      "rotation_invariant_error_deg": 0.7336960316225287,
-      "reverse_translation_m": 0.013565042721589003,
-      "reverse_rotation_deg": 0.32552304340992394,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio",
-        "heldout_rmse"
-      ]
-    },
-    {
-      "i": 11,
-      "j": 12,
-      "heldout_inlier_ratio": 0.650555275113579,
-      "heldout_inlier_rmse_m": 0.12104882943540898,
-      "rotation_invariant_error_deg": 0.16828520325654495,
-      "reverse_translation_m": 0.003583199374507776,
-      "reverse_rotation_deg": 0.03300207707174117,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio"
-      ]
-    },
-    {
-      "i": 11,
-      "j": 13,
-      "heldout_inlier_ratio": 0.5698054068172914,
-      "heldout_inlier_rmse_m": 0.1285866274322162,
-      "rotation_invariant_error_deg": 0.7985328447719269,
-      "reverse_translation_m": 0.016001435752898144,
-      "reverse_rotation_deg": 0.11059625579950419,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio",
-        "rotation_conjugacy_invariant"
-      ]
-    },
-    {
-      "i": 11,
-      "j": 14,
-      "heldout_inlier_ratio": 0.6420881321982974,
-      "heldout_inlier_rmse_m": 0.12414062948335593,
-      "rotation_invariant_error_deg": 0.9364144322988963,
-      "reverse_translation_m": 0.012369101516230316,
-      "reverse_rotation_deg": 0.01870506040605898,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio",
-        "rotation_conjugacy_invariant"
-      ]
-    },
-    {
-      "i": 12,
-      "j": 13,
-      "heldout_inlier_ratio": 0.6972966112450819,
-      "heldout_inlier_rmse_m": 0.1168089621311632,
-      "rotation_invariant_error_deg": 0.9873386907551662,
-      "reverse_translation_m": 0.01141023876783206,
-      "reverse_rotation_deg": 0.04779713993428384,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio",
-        "rotation_conjugacy_invariant"
-      ]
-    },
-    {
-      "i": 12,
-      "j": 14,
-      "heldout_inlier_ratio": 0.8749086479902558,
-      "heldout_inlier_rmse_m": 0.09521299965540625,
-      "rotation_invariant_error_deg": 0.7228372421049902,
-      "reverse_translation_m": 0.006159472215773367,
-      "reverse_rotation_deg": 0.014929948455569534,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 12,
-      "j": 15,
-      "heldout_inlier_ratio": 0.7022030893897189,
-      "heldout_inlier_rmse_m": 0.11797995277580106,
-      "rotation_invariant_error_deg": 0.34088274652411243,
-      "reverse_translation_m": 0.008495008365046321,
-      "reverse_rotation_deg": 0.10278299144703042,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 13,
-      "j": 14,
-      "heldout_inlier_ratio": 0.6955810147299509,
-      "heldout_inlier_rmse_m": 0.11455956122705321,
-      "rotation_invariant_error_deg": 1.7562842773469676,
-      "reverse_translation_m": 0.01022866463344607,
-      "reverse_rotation_deg": 0.03515966054944392,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio",
-        "rotation_conjugacy_invariant"
-      ]
-    },
-    {
-      "i": 13,
-      "j": 15,
-      "heldout_inlier_ratio": 0.868300353819945,
-      "heldout_inlier_rmse_m": 0.1045421356808481,
-      "rotation_invariant_error_deg": 0.6514660625063202,
-      "reverse_translation_m": 0.002375025382773949,
-      "reverse_rotation_deg": 0.01072504764419793,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 13,
-      "j": 16,
-      "heldout_inlier_ratio": 0.7265456392027422,
-      "heldout_inlier_rmse_m": 0.1096252966406241,
-      "rotation_invariant_error_deg": 0.48470120712195097,
-      "reverse_translation_m": 0.008958927594996346,
-      "reverse_rotation_deg": 0.03041438512426757,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 14,
-      "j": 15,
-      "heldout_inlier_ratio": 0.7120070334086913,
-      "heldout_inlier_rmse_m": 0.11868441290330703,
-      "rotation_invariant_error_deg": 1.1046128571505562,
-      "reverse_translation_m": 0.002571958018980028,
-      "reverse_rotation_deg": 0.055069197511519646,
-      "accepted": false,
-      "rejection_reasons": [
-        "rotation_conjugacy_invariant"
-      ]
-    },
-    {
-      "i": 14,
-      "j": 16,
-      "heldout_inlier_ratio": 0.5918615984405458,
-      "heldout_inlier_rmse_m": 0.12229328437386515,
-      "rotation_invariant_error_deg": 1.240484202449963,
-      "reverse_translation_m": 0.014273957859025563,
-      "reverse_rotation_deg": 0.25325650727957555,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio",
-        "rotation_conjugacy_invariant"
-      ]
-    },
-    {
-      "i": 14,
-      "j": 17,
-      "heldout_inlier_ratio": 0.6694009445687298,
-      "heldout_inlier_rmse_m": 0.12443900216431925,
-      "rotation_invariant_error_deg": 0.4822313558358869,
-      "reverse_translation_m": 0.0178952010004604,
-      "reverse_rotation_deg": 0.10920228290609924,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio"
-      ]
-    },
-    {
-      "i": 15,
-      "j": 16,
-      "heldout_inlier_ratio": 0.702887537993921,
-      "heldout_inlier_rmse_m": 0.11495230769293859,
-      "rotation_invariant_error_deg": 0.16340661916523302,
-      "reverse_translation_m": 0.013545291843396808,
-      "reverse_rotation_deg": 0.03346625178333291,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 15,
-      "j": 17,
-      "heldout_inlier_ratio": 0.7429531936901991,
-      "heldout_inlier_rmse_m": 0.11679524427533863,
-      "rotation_invariant_error_deg": 0.050851974711708436,
-      "reverse_translation_m": 0.009894110027911674,
-      "reverse_rotation_deg": 0.07786907370565768,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 15,
-      "j": 18,
-      "heldout_inlier_ratio": 0.7209645010046886,
-      "heldout_inlier_rmse_m": 0.11716134583909138,
-      "rotation_invariant_error_deg": 0.26477152026833295,
-      "reverse_translation_m": 0.01165472931184747,
-      "reverse_rotation_deg": 0.13683586564190106,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 16,
-      "j": 17,
-      "heldout_inlier_ratio": 0.618922305764411,
-      "heldout_inlier_rmse_m": 0.11254196340940027,
-      "rotation_invariant_error_deg": 0.05549924909794868,
-      "reverse_translation_m": 0.031047860213503222,
-      "reverse_rotation_deg": 0.10375098145236057,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio"
-      ]
-    },
-    {
-      "i": 16,
-      "j": 18,
-      "heldout_inlier_ratio": 0.6890156918687589,
-      "heldout_inlier_rmse_m": 0.11084024896736888,
-      "rotation_invariant_error_deg": 0.06628387310017914,
-      "reverse_translation_m": 0.01556225520373068,
-      "reverse_rotation_deg": 0.02495881796885156,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio"
-      ]
-    },
-    {
-      "i": 16,
-      "j": 19,
-      "heldout_inlier_ratio": 0.8685060899826,
-      "heldout_inlier_rmse_m": 0.10135543575024479,
-      "rotation_invariant_error_deg": 0.3152201090321931,
-      "reverse_translation_m": 0.0029522513838272538,
-      "reverse_rotation_deg": 0.02753369989558306,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 17,
-      "j": 18,
-      "heldout_inlier_ratio": 0.7697708305735859,
-      "heldout_inlier_rmse_m": 0.10648049893472189,
-      "rotation_invariant_error_deg": 0.2007356143443495,
-      "reverse_translation_m": 0.010582276181446961,
-      "reverse_rotation_deg": 0.03959905194910384,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 17,
-      "j": 19,
-      "heldout_inlier_ratio": 0.6293759512937596,
-      "heldout_inlier_rmse_m": 0.10954497717502036,
-      "rotation_invariant_error_deg": 0.25455213438726787,
-      "reverse_translation_m": 0.012240937390578075,
-      "reverse_rotation_deg": 0.06030618467886912,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio"
-      ]
-    },
-    {
-      "i": 18,
-      "j": 19,
-      "heldout_inlier_ratio": 0.6827314510833881,
-      "heldout_inlier_rmse_m": 0.10834806615100022,
-      "rotation_invariant_error_deg": 0.4543752098998439,
-      "reverse_translation_m": 0.012418496053909043,
-      "reverse_rotation_deg": 0.11905018881087433,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio"
-      ]
-    },
-    {
-      "i": 19,
-      "j": 20,
-      "heldout_inlier_ratio": 0.6234734541714874,
-      "heldout_inlier_rmse_m": 0.1277530580405749,
-      "rotation_invariant_error_deg": 0.038058886117710244,
-      "reverse_translation_m": 0.009237066767190974,
-      "reverse_rotation_deg": 0.22540488888103255,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio"
-      ]
-    },
-    {
-      "i": 20,
-      "j": 21,
-      "heldout_inlier_ratio": 0.6596992097884272,
-      "heldout_inlier_rmse_m": 0.12147955429086157,
-      "rotation_invariant_error_deg": 0.33308699073501913,
-      "reverse_translation_m": 0.014725537493639118,
-      "reverse_rotation_deg": 0.19884224722867958,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio"
-      ]
-    },
-    {
-      "i": 20,
-      "j": 22,
-      "heldout_inlier_ratio": 0.5739414499308958,
-      "heldout_inlier_rmse_m": 0.13255415786946775,
-      "rotation_invariant_error_deg": 0.49226728584466883,
-      "reverse_translation_m": 0.006040617548523686,
-      "reverse_rotation_deg": 0.19706632354686843,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio",
-        "heldout_rmse"
-      ]
-    },
-    {
-      "i": 20,
-      "j": 23,
-      "heldout_inlier_ratio": 0.6456945156330087,
-      "heldout_inlier_rmse_m": 0.12075756038041646,
-      "rotation_invariant_error_deg": 0.4097857444953803,
-      "reverse_translation_m": 0.02294901101348451,
-      "reverse_rotation_deg": 0.19112109479244785,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio"
-      ]
-    },
-    {
-      "i": 21,
-      "j": 22,
-      "heldout_inlier_ratio": 0.7716237647919971,
-      "heldout_inlier_rmse_m": 0.1163227135854156,
-      "rotation_invariant_error_deg": 0.20299476740399935,
-      "reverse_translation_m": 0.009963578798274064,
-      "reverse_rotation_deg": 0.1533289219532281,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 21,
-      "j": 23,
-      "heldout_inlier_ratio": 0.8576224819696593,
-      "heldout_inlier_rmse_m": 0.09865676260631218,
-      "rotation_invariant_error_deg": 0.08196486868098773,
-      "reverse_translation_m": 0.00392257332509571,
-      "reverse_rotation_deg": 0.00898487649366286,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 21,
-      "j": 24,
-      "heldout_inlier_ratio": 0.7715940569126165,
-      "heldout_inlier_rmse_m": 0.11361504553552869,
-      "rotation_invariant_error_deg": 0.4941752812908575,
-      "reverse_translation_m": 0.007475673605592584,
-      "reverse_rotation_deg": 0.04227769001294981,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 22,
-      "j": 23,
-      "heldout_inlier_ratio": 0.7396689147762109,
-      "heldout_inlier_rmse_m": 0.11702962848624102,
-      "rotation_invariant_error_deg": 0.07221906147914936,
-      "reverse_translation_m": 0.018140159434305195,
-      "reverse_rotation_deg": 0.1184732181610567,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 22,
-      "j": 25,
-      "heldout_inlier_ratio": 0.736861094407697,
-      "heldout_inlier_rmse_m": 0.1158639471011007,
-      "rotation_invariant_error_deg": 0.19913456977445776,
-      "reverse_translation_m": 0.01855955252477569,
-      "reverse_rotation_deg": 0.06452602797902846,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 23,
-      "j": 24,
-      "heldout_inlier_ratio": 0.7853164556962026,
-      "heldout_inlier_rmse_m": 0.11287254423109305,
-      "rotation_invariant_error_deg": 0.5709657470221714,
-      "reverse_translation_m": 0.002195759849349955,
-      "reverse_rotation_deg": 0.03211295915781923,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 23,
-      "j": 25,
-      "heldout_inlier_ratio": 0.6881127450980392,
-      "heldout_inlier_rmse_m": 0.12303206700403985,
-      "rotation_invariant_error_deg": 2.3734594630006685,
-      "reverse_translation_m": 0.004939188019097873,
-      "reverse_rotation_deg": 0.12964064637099942,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio",
-        "rotation_conjugacy_invariant"
-      ]
-    },
-    {
-      "i": 23,
-      "j": 26,
-      "heldout_inlier_ratio": 0.6852618757612667,
-      "heldout_inlier_rmse_m": 0.12040544972155913,
-      "rotation_invariant_error_deg": 0.15748268923838538,
-      "reverse_translation_m": 0.006032445250251169,
-      "reverse_rotation_deg": 0.14039335223022864,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio"
-      ]
-    },
-    {
-      "i": 24,
-      "j": 25,
-      "heldout_inlier_ratio": 0.677667493796526,
-      "heldout_inlier_rmse_m": 0.1234631580581415,
-      "rotation_invariant_error_deg": 0.10001483417664048,
-      "reverse_translation_m": 0.023479226891170584,
-      "reverse_rotation_deg": 0.16273859629355136,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio"
-      ]
-    },
-    {
-      "i": 24,
-      "j": 26,
-      "heldout_inlier_ratio": 0.6530209617755857,
-      "heldout_inlier_rmse_m": 0.12710147612984235,
-      "rotation_invariant_error_deg": 0.7149536911782421,
-      "reverse_translation_m": 0.013311199903813952,
-      "reverse_rotation_deg": 0.18193579850150715,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio"
-      ]
-    },
-    {
-      "i": 24,
-      "j": 27,
-      "heldout_inlier_ratio": 0.6649014778325123,
-      "heldout_inlier_rmse_m": 0.12481159614427853,
-      "rotation_invariant_error_deg": 0.5126580224867077,
-      "reverse_translation_m": 0.01787282120544869,
-      "reverse_rotation_deg": 0.1680556511623414,
-      "accepted": false,
-      "rejection_reasons": [
-        "overlap_ratio"
-      ]
-    },
-    {
-      "i": 25,
-      "j": 26,
-      "heldout_inlier_ratio": 0.9127837514934289,
-      "heldout_inlier_rmse_m": 0.09581429165893823,
-      "rotation_invariant_error_deg": 0.0540123863768045,
-      "reverse_translation_m": 0.0035341488401767693,
-      "reverse_rotation_deg": 0.018440169300164816,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 25,
-      "j": 27,
-      "heldout_inlier_ratio": 0.8510739856801909,
-      "heldout_inlier_rmse_m": 0.10418150333394123,
-      "rotation_invariant_error_deg": 0.24124343828050598,
-      "reverse_translation_m": 0.006103232696003387,
-      "reverse_rotation_deg": 0.13366555345654282,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 25,
-      "j": 28,
-      "heldout_inlier_ratio": 0.8853518429870751,
-      "heldout_inlier_rmse_m": 0.10761251229651997,
-      "rotation_invariant_error_deg": 0.31032271091986274,
-      "reverse_translation_m": 0.0034832316659127254,
-      "reverse_rotation_deg": 0.03697744201993421,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 26,
-      "j": 27,
-      "heldout_inlier_ratio": 0.9183867141162515,
-      "heldout_inlier_rmse_m": 0.09074909570650827,
-      "rotation_invariant_error_deg": 0.1549242409220426,
-      "reverse_translation_m": 0.00040199505815399084,
-      "reverse_rotation_deg": 0.011554314408123986,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 26,
-      "j": 28,
-      "heldout_inlier_ratio": 0.8853200095170116,
-      "heldout_inlier_rmse_m": 0.1060418493965508,
-      "rotation_invariant_error_deg": 0.2535175622408232,
-      "reverse_translation_m": 0.007320240476184076,
-      "reverse_rotation_deg": 0.12930335868606002,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 26,
-      "j": 29,
-      "heldout_inlier_ratio": 0.7880466815984911,
-      "heldout_inlier_rmse_m": 0.10871274240898261,
-      "rotation_invariant_error_deg": 0.6273264323501451,
-      "reverse_translation_m": 0.006395243061058376,
-      "reverse_rotation_deg": 0.039517035902872893,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 27,
-      "j": 28,
-      "heldout_inlier_ratio": 0.9289448669201521,
-      "heldout_inlier_rmse_m": 0.08505095515326752,
-      "rotation_invariant_error_deg": 0.06642597183531862,
-      "reverse_translation_m": 0.005157655135593023,
-      "reverse_rotation_deg": 0.02266680787731125,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 27,
-      "j": 29,
-      "heldout_inlier_ratio": 0.7872365477452019,
-      "heldout_inlier_rmse_m": 0.1045551346483567,
-      "rotation_invariant_error_deg": 0.46617350279205994,
-      "reverse_translation_m": 0.011650639607012715,
-      "reverse_rotation_deg": 0.16247191340775555,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 28,
-      "j": 29,
-      "heldout_inlier_ratio": 0.8000944621560987,
-      "heldout_inlier_rmse_m": 0.10865475473581254,
-      "rotation_invariant_error_deg": 0.3806425437120282,
-      "reverse_translation_m": 0.001139416496730335,
-      "reverse_rotation_deg": 0.02119627748722763,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 28,
-      "j": 30,
-      "heldout_inlier_ratio": 0.7721000935453695,
-      "heldout_inlier_rmse_m": 0.10815845248211022,
-      "rotation_invariant_error_deg": 0.37744105649966286,
-      "reverse_translation_m": 0.004462522533166182,
-      "reverse_rotation_deg": 0.03583195432667579,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 28,
-      "j": 31,
-      "heldout_inlier_ratio": 0.7465330381074466,
-      "heldout_inlier_rmse_m": 0.10669666534392452,
-      "rotation_invariant_error_deg": 0.22580828656680296,
-      "reverse_translation_m": 0.0063576490595817345,
-      "reverse_rotation_deg": 0.04001661521442668,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 30,
-      "j": 31,
-      "heldout_inlier_ratio": 0.8185562292643862,
-      "heldout_inlier_rmse_m": 0.09615795732951272,
-      "rotation_invariant_error_deg": 0.608920703223724,
-      "reverse_translation_m": 0.0049991634555749285,
-      "reverse_rotation_deg": 0.021384795873435915,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 30,
-      "j": 32,
-      "heldout_inlier_ratio": 0.8041343079031521,
-      "heldout_inlier_rmse_m": 0.09953343153684206,
-      "rotation_invariant_error_deg": 0.7410181887106262,
-      "reverse_translation_m": 0.0031713764951448154,
-      "reverse_rotation_deg": 0.023767377748422268,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 31,
-      "j": 32,
-      "heldout_inlier_ratio": 0.8157085941946499,
-      "heldout_inlier_rmse_m": 0.09925944770258255,
-      "rotation_invariant_error_deg": 0.1261576421594981,
-      "reverse_translation_m": 0.004719596234885736,
-      "reverse_rotation_deg": 0.06146741991683861,
-      "accepted": true,
-      "rejection_reasons": []
-    },
-    {
-      "i": 31,
-      "j": 33,
-      "heldout_inlier_ratio": 0.760345235280208,
-      "heldout_inlier_rmse_m": 0.09698073659477859,
-      "rotation_invariant_error_deg": 0.11718235773108177,
-      "reverse_translation_m": 0.0019366659831763946,
-      "reverse_rotation_deg": 0.026538404789073603,
-      "accepted": true,
-      "rejection_reasons": []
-    }
-  ]
-}
@@ -1,346 +0,0 @@
-{
-  "schema_version": 1,
-  "success": true,
-  "convention": "T_RTK_lidar maps raw LiDAR points into the RTK navigation frame",
-  "equation": "A_RTK_ij X = X B_LiDAR_ij",
-  "frames": {
-    "RTK": {
-      "origin": "GGA positioning reference point; confirm ANT1/reference antenna in receiver configuration",
-      "x_axis": "horizontal projection of the rawHeading baseline direction reported by the receiver",
-      "y_axis": "left",
-      "z_axis": "up",
-      "yaw_enu_deg": "90 - rawHeadingDeg"
-    },
-    "LiDAR": "raw LiDAR sensor frame"
-  },
-  "backend": "open3d_gicp",
-  "measured_lidar_extrinsic_used_as_initial": false,
-  "body_heading_offset_used": false,
-  "body_antenna_lever_xy_used": false,
-  "translation_m": [
-    1.6368478986289665,
-    -0.2424214446279586,
-    0.08408840390370595
-  ],
-  "rotation_rpy_deg_xyz": [
-    -0.8529068506201997,
-    1.3285763243955087,
-    -22.150129507458946
-  ],
-  "quaternion_xyzw": [
-    -0.005076802391609619,
-    0.012807178841486907,
-    -0.19199197181185435,
-    0.9812997936448346
-  ],
-  "matrix_4x4": [
-    [
-      0.9259501178598366,
-      0.3766733256085306,
-      0.027084774511975725,
-      1.6368478986289665
-    ],
-    [
-      -0.3769334036732196,
-      0.9262266176745453,
-      0.005045979240275979,
-      -0.2424214446279586
-    ],
-    [
-      -0.023185953305318648,
-      -0.014881481316772506,
-      0.9996204044951974,
-      0.08408840390370595
-    ],
-    [
-      0.0,
-      0.0,
-      0.0,
-      1.0
-    ]
-  ],
-  "quality": {
-    "stations": 34,
-    "pairs": 41,
-    "residuals": {
-      "pairs": 41,
-      "translation_m": {
-        "rms": 0.11160362192890017,
-        "median": 0.06165368731012592,
-        "p90": 0.12167778728549836,
-        "p95": 0.12315870952465978,
-        "max": 0.45797693508387505
-      },
-      "rotation_deg": {
-        "rms": 1.082945057501201,
-        "median": 0.744916302248719,
-        "p90": 1.523342607477641,
-        "p95": 1.864505235125524,
-        "max": 4.332008025818138
-      },
-      "per_pair": [
-        {
-          "pair_index": 0,
-          "translation_m": 0.03696060713715898,
-          "rotation_deg": 0.7820683861381595
-        },
-        {
-          "pair_index": 1,
-          "translation_m": 0.039028552931951836,
-          "rotation_deg": 0.6949685743890126
-        },
-        {
-          "pair_index": 2,
-          "translation_m": 0.01732676228513485,
-          "rotation_deg": 1.0438060827136844
-        },
-        {
-          "pair_index": 3,
-          "translation_m": 0.0574978038017053,
-          "rotation_deg": 0.394946532754576
-        },
-        {
-          "pair_index": 4,
-          "translation_m": 0.10801668349596734,
-          "rotation_deg": 0.29336064351197005
-        },
-        {
-          "pair_index": 5,
-          "translation_m": 0.06577881887454434,
-          "rotation_deg": 0.5542732613092561
-        },
-        {
-          "pair_index": 6,
-          "translation_m": 0.06363940755916121,
-          "rotation_deg": 0.5890545170693674
-        },
-        {
-          "pair_index": 7,
-          "translation_m": 0.0640679503495012,
-          "rotation_deg": 0.746120063911557
-        },
-        {
-          "pair_index": 8,
-          "translation_m": 0.021009631897429052,
-          "rotation_deg": 0.6663196620233134
-        },
-        {
-          "pair_index": 9,
-          "translation_m": 0.08609573827254784,
-          "rotation_deg": 0.3687772861174946
-        },
-        {
-          "pair_index": 10,
-          "translation_m": 0.02732304440983888,
-          "rotation_deg": 0.5222061823280917
-        },
-        {
-          "pair_index": 11,
-          "translation_m": 0.04489317903985948,
-          "rotation_deg": 0.8700136243621963
-        },
-        {
-          "pair_index": 12,
-          "translation_m": 0.07260882213085644,
-          "rotation_deg": 0.463182802330913
-        },
-        {
-          "pair_index": 13,
-          "translation_m": 0.08831772542234817,
-          "rotation_deg": 0.659009021415606
-        },
-        {
-          "pair_index": 14,
-          "translation_m": 0.053804778572191785,
-          "rotation_deg": 1.864505235125524
-        },
-        {
-          "pair_index": 15,
-          "translation_m": 0.12315870952465978,
-          "rotation_deg": 1.9856861904193648
-        },
-        {
-          "pair_index": 16,
-          "translation_m": 0.06880647181572622,
-          "rotation_deg": 0.27331219092746517
-        },
-        {
-          "pair_index": 17,
-          "translation_m": 0.059582300078982915,
-          "rotation_deg": 0.3410200031868246
-        },
-        {
-          "pair_index": 18,
-          "translation_m": 0.034895300030885576,
-          "rotation_deg": 0.5283410202190479
-        },
-        {
-          "pair_index": 19,
-          "translation_m": 0.002366788176226253,
-          "rotation_deg": 0.3852038090000981
-        },
-        {
-          "pair_index": 20,
-          "translation_m": 0.06233663305653708,
-          "rotation_deg": 0.6208000110952433
-        },
-        {
-          "pair_index": 21,
-          "translation_m": 0.012626372544105475,
-          "rotation_deg": 0.38292733656142575
-        },
-        {
-          "pair_index": 22,
-          "translation_m": 0.047494831014178666,
-          "rotation_deg": 1.7712490998072288
-        },
-        {
-          "pair_index": 23,
-          "translation_m": 0.06165368731012592,
-          "rotation_deg": 0.8670376638275039
-        },
-        {
-          "pair_index": 24,
-          "translation_m": 0.12172160687143056,
-          "rotation_deg": 4.332008025818138
-        },
-        {
-          "pair_index": 25,
-          "translation_m": 0.06185808030991236,
-          "rotation_deg": 1.523342607477641
-        },
-        {
-          "pair_index": 26,
-          "translation_m": 0.049450564075379046,
-          "rotation_deg": 0.9499016191505171
-        },
-        {
-          "pair_index": 27,
-          "translation_m": 0.016599648000378425,
-          "rotation_deg": 0.7685476074882036
-        },
-        {
-          "pair_index": 28,
-          "translation_m": 0.028501035078080036,
-          "rotation_deg": 0.9107236146504971
-        },
-        {
-          "pair_index": 29,
-          "translation_m": 0.05418062512497841,
-          "rotation_deg": 0.49604339332524483
-        },
-        {
-          "pair_index": 30,
-          "translation_m": 0.02534272431328254,
-          "rotation_deg": 0.2798196345939902
-        },
-        {
-          "pair_index": 31,
-          "translation_m": 0.1006162178003358,
-          "rotation_deg": 0.9066113005741254
-        },
-        {
-          "pair_index": 32,
-          "translation_m": 0.023348232889945957,
-          "rotation_deg": 0.29316028754024664
-        },
-        {
-          "pair_index": 33,
-          "translation_m": 0.06352851658442724,
-          "rotation_deg": 1.123276625138155
-        },
-        {
-          "pair_index": 34,
-          "translation_m": 0.06685972566531029,
-          "rotation_deg": 0.7607492800980359
-        },
-        {
-          "pair_index": 35,
-          "translation_m": 0.3534873096430149,
-          "rotation_deg": 0.744916302248719
-        },
-        {
-          "pair_index": 36,
-          "translation_m": 0.45797693508387505,
-          "rotation_deg": 0.7684054290628693
-        },
-        {
-          "pair_index": 37,
-          "translation_m": 0.10414608850682983,
-          "rotation_deg": 0.8325040370564112
-        },
-        {
-          "pair_index": 38,
-          "translation_m": 0.11058068261532474,
-          "rotation_deg": 0.8695456609442757
-        },
-        {
-          "pair_index": 39,
-          "translation_m": 0.03405996353456899,
-          "rotation_deg": 0.7913171839743647
-        },
-        {
-          "pair_index": 40,
-          "translation_m": 0.12167778728549836,
-          "rotation_deg": 0.16247416554150054
-        }
-      ]
-    },
-    "weighted_jacobian_condition_number": 7.006062559882809,
-    "linearized_one_sigma": {
-      "translation_m": [
-        0.007129000573825617,
-        0.007213998960440989,
-        0.0046717381822970125
-      ],
-      "rotation_deg": [
-        0.06695429523124989,
-        0.06503891988826449,
-        0.14115095959507304
-      ],
-      "warning": "conditional local estimate; bootstrap is the primary stability check"
-    },
-    "bootstrap": {
-      "runs": 100,
-      "order": [
-        "x_m",
-        "y_m",
-        "z_m",
-        "roll_deg",
-        "pitch_deg",
-        "yaw_deg"
-      ],
-      "std": [
-        0.002608526149658222,
-        0.0028276716521100863,
-        0.0023368456176917846,
-        0.0795227773764157,
-        0.07300189168975164,
-        0.10947642324382982
-      ],
-      "p025": [
-        1.6327841703535608,
-        -0.24653273132861378,
-        0.07991981050403614,
-        -0.9872072161381827,
-        1.1864367848117605,
-        -22.364052444549895
-      ],
-      "p975": [
-        1.6431149050874008,
-        -0.2355422823589496,
-        0.08894657837563841,
-        -0.6759368885907463,
-        1.4669998309298398,
-        -21.953029139092944
-      ]
-    }
-  },
-  "z_constraint": {
-    "observable_from_planar_AX_XB": false,
-    "method": "LiDAR ground planes plus externally supplied RTK reference-point height above ground",
-    "rtk_reference_height_above_ground_m": 0.8535,
-    "warning": "z is conditional on the supplied RTK antenna height; it is not independently identified by planar Ackermann motion"
-  },
-  "important_limit": "AX residual and bootstrap quantify internal consistency, not independent centimetre-grade absolute certification"
-}
@@ -1,156 +0,0 @@
-i,j,rtk_translation_m,rtk_rotation_deg,heldout_inlier_ratio,heldout_inlier_rmse_m,hessian_rank,hessian_condition,reverse_translation_m,reverse_rotation_deg,multistart_success_rate,accepted,rejection_reasons
-0,1,1.6721594124489447,24.171297449440814,0.8193962748876044,0.11049675306366954,6,14.013392649694936,0.026679665410762447,0.12399936190197167,1.0,True,
-0,2,2.0412175279332088,80.09074797031303,0.7525388867463684,0.11492533799491883,6,14.962108606929117,0.0018464756299783867,0.03093241101186373,1.0,True,
-0,3,6.30529961936688,79.9158388329924,0.628093901505486,0.12365050970311502,6,23.490589415548122,0.010524333328230958,0.23898233567764737,0.5,True,
-0,4,2.238422471863255,130.25781950514033,0.693351593625498,0.11485738628517483,6,16.305124521706706,0.003627491377787396,0.0495829610363469,1.0,True,
-0,5,7.5269579262553155,88.82606603101335,0.6015065913370998,0.12959859333012305,6,26.27293180169831,0.013710015050868782,0.26025123892797375,1.0,True,
-1,2,1.2843386040405174,55.9194505208722,0.7981310803891449,0.11167434332282765,6,13.484382276710306,0.0507423684848027,0.517099810570953,1.0,False,forward_reverse_rotation
-1,3,5.433888607887495,55.744541383551606,0.678820988438572,0.12109867185657658,6,27.478714016210855,0.008324315958294127,0.2568985217684905,1.0,True,
-1,4,1.5299424710613851,106.08652205569952,0.6772473651580905,0.1115749218038809,6,13.250326799303036,3.109603769112268,6.665689673243039,1.0,False,forward_reverse_translation;forward_reverse_rotation
-1,5,7.072560501394692,64.65476858157254,0.618779694923731,0.1278696355679149,6,21.376356908960656,0.012044684321696756,0.5185219918090542,1.0,False,forward_reverse_rotation
-1,6,8.049825623399226,8.10898363252498,0.02911760982402836,0.16814699881028172,6,51.15739724954147,2.3526085660101135,12.54864815902537,0.0,False,backend_not_converged;heldout_inlier_ratio;heldout_inlier_rmse;forward_reverse_translation;forward_reverse_rotation;multistart_instability
-2,3,4.340252832276203,0.1749091373206093,0.7609663064208518,0.11583878636902087,6,13.299472485717942,0.007842434748069874,0.017111535671962216,1.0,True,
-2,4,0.2520257253555564,50.1670715348273,0.796748976299789,0.1148434235904791,6,12.041666425070192,0.011210942709506708,0.11462542679279858,1.0,True,
-2,5,5.8286926576588955,8.735318060700322,0.7112112112112112,0.12001318292058727,6,15.618628103418056,0.011299298862678088,0.02158353743310138,1.0,True,
-2,6,6.928094074716812,47.810466888347236,0.058659571772456606,0.1689308471089219,6,22.346184538451386,2.376212271528816,4.191297741726165,0.0,False,heldout_inlier_ratio;heldout_inlier_rmse;forward_reverse_translation;forward_reverse_rotation;multistart_instability
-2,7,6.405228405426323,73.1220331951562,0.05777324320877439,0.16763553481687163,6,11.228331216247241,2.3764975144091216,1.7142235592052535,0.0,False,heldout_inlier_ratio;heldout_inlier_rmse;forward_reverse_translation;forward_reverse_rotation;multistart_instability
-3,4,4.1070657333447205,50.34198067214791,0.6957378664695738,0.11619002437046365,6,17.76967737811838,3.222808271854619,15.590678196925943,1.0,False,forward_reverse_translation;forward_reverse_rotation
-3,5,2.2886212019715484,8.910227198020936,0.7989069680784996,0.10775905778143813,6,12.151555874812614,0.013532537604982006,0.06485728954506689,1.0,True,
-3,6,2.618668779147775,47.63555775102663,0.8435613682092555,0.11222309863990189,6,13.914901775514537,0.00951276519885504,0.09743636874086448,1.0,True,
-3,7,2.7963089245830335,72.9471240578356,0.8651898734177215,0.11184568072686091,6,12.975777248765162,0.010827690226297664,0.12520280777371842,1.0,True,
-3,8,2.6983089912812726,163.5692883655715,0.825590155700653,0.1078798726707026,6,12.959410142765158,0.005608807919239624,0.021536601402144962,1.0,True,
-4,5,5.5767898078953735,41.431753474126985,0.6652516676773802,0.12444359189600171,6,17.028714587649738,0.0031552835887398907,0.05824661275042354,1.0,True,
-4,6,6.68811709459501,97.97753842317455,0.03891480481217775,0.16130442983218543,6,53.12725754116488,5.386834276571879,3.248443974085464,1.0,False,heldout_inlier_ratio;heldout_inlier_rmse;forward_reverse_translation;forward_reverse_rotation
-4,7,6.153247042777442,123.28910472998353,0.01717321472695824,0.16597934102353687,6,197.50040966862915,1.8946664283973234,13.38395621317346,0.0,False,heldout_inlier_ratio;heldout_inlier_rmse;forward_reverse_translation;forward_reverse_rotation;multistart_instability
-4,8,6.802787549686365,146.08873096228075,0.7129198332924737,0.11349036082807593,6,18.616925126379055,4.094017521116411,1.0439112418969816,1.0,False,forward_reverse_translation;forward_reverse_rotation
-4,9,3.018117089902297,158.11607430100386,0.3323991714390155,0.13218238008205907,6,82.91588951856485,0.17404817666776692,0.7491904391974432,1.0,False,heldout_inlier_ratio;forward_reverse_translation;forward_reverse_rotation
-5,6,2.0562758992403367,56.545784949047565,0.7604901596732269,0.11101745020162715,6,16.790595774247244,0.01028831511886355,0.031120237309440944,1.0,True,
-5,7,0.5812996709001959,81.85735125585653,0.8092687180764918,0.10777871969807898,6,15.203386410549202,0.010579733674272045,0.033334626234333836,1.0,True,
-5,8,2.6887856969568644,172.47951556359513,0.40909652700531457,0.13021927164196687,6,88.03642906190348,2.282274682790372,1.1841285568948536,1.0,False,forward_reverse_translation;forward_reverse_rotation
-5,9,7.501939677383991,160.45217222486954,0.3361179361179361,0.13367797847230906,6,83.15270070161475,5.255898884183195,7.177742907146106,0.5,False,heldout_inlier_ratio;forward_reverse_translation;forward_reverse_rotation
-5,10,7.507648391453363,143.54232919109307,0.5442391832766165,0.13173165083201846,6,26.337222871823244,4.175333583746659,14.054946454556381,1.0,False,forward_reverse_translation;forward_reverse_rotation
-6,7,1.9723544820714844,25.311566306808967,0.8539354187689203,0.10462253085152279,6,12.50291076661203,0.0028433142784691904,0.028424505435071433,1.0,True,
-6,8,0.7288530799256238,115.93373061454484,0.7757757757757757,0.11316400028465075,6,15.521747346102574,0.007224674181692249,0.10395594559619384,1.0,True,
-6,9,7.898758206937296,103.90638727582184,0.6009202835468226,0.1259448851291812,6,28.795189977892598,4.270523553799638,1.1342776748452013,0.5,False,forward_reverse_translation;forward_reverse_rotation
-6,10,8.382165572419371,159.91188585985944,0.048837495386886455,0.16976022756819517,6,33.03889916791793,8.842801960353667,9.008466157678757,0.0,False,heldout_inlier_ratio;heldout_inlier_rmse;forward_reverse_translation;forward_reverse_rotation;multistart_instability
-6,11,11.12084001821623,138.54043874549885,0.026144624410151765,0.1725705028585339,6,71.57462790292871,2.4153309235424008,10.273490365819672,0.0,False,heldout_inlier_ratio;heldout_inlier_rmse;forward_reverse_translation;forward_reverse_rotation;multistart_instability
-7,8,2.6747845281541447,90.62216430773587,0.8340050377833753,0.11132245152738934,6,16.17035077702852,0.004806949653405576,0.020340398656013788,1.0,True,
-7,9,8.06955581661561,78.59482096901287,0.6160971335586432,0.12637042722943573,6,24.178664977065605,4.367245914434154,12.796681469446304,1.0,False,forward_reverse_translation;forward_reverse_rotation
-7,10,8.088675434881791,134.6003195530505,0.04890429614956048,0.17569950505457485,6,17.12714594442953,9.4426170895353,17.952544168886714,0.0,False,backend_not_converged;heldout_inlier_ratio;heldout_inlier_rmse;forward_reverse_translation;forward_reverse_rotation;multistart_instability
-7,11,10.488793105910844,113.2288724386899,0.03723199383746309,0.16334679446297104,6,88.61274561425562,7.5280305092863244,55.909263646297305,0.0,False,heldout_inlier_ratio;heldout_inlier_rmse;forward_reverse_translation;forward_reverse_rotation;multistart_instability
-7,12,13.79542539595065,94.95104818613552,0.023342903507676944,0.1706723033660903,6,71.17833596146563,5.060797550499477,27.755426788516992,0.0,False,heldout_inlier_ratio;heldout_inlier_rmse;forward_reverse_translation;forward_reverse_rotation;multistart_instability
-8,9,7.73033999229505,12.027343338722995,0.6407549981373402,0.12189583104066957,6,28.06713348388492,2.3076573667891433,3.487290408239611,1.0,False,forward_reverse_translation;forward_reverse_rotation
-8,10,8.378411682322204,43.97815524531458,0.6075420709986488,0.12933255488441212,6,20.429633989572718,4.865121579871581,2.927465091074779,1.0,False,forward_reverse_translation;forward_reverse_rotation
-8,11,11.214115106391473,22.60670813095403,0.03922067999490641,0.16246238376196148,6,54.700772476792416,2.594876837596059,7.9735312612345846,0.0,False,heldout_inlier_ratio;heldout_inlier_rmse;forward_reverse_translation;forward_reverse_rotation;multistart_instability
-8,12,14.373410961212507,4.328883878399634,0.023529411764705882,0.1707990278738782,6,65.92814703722017,0.7170106443431138,1.1046762967347212,0.5,False,heldout_inlier_ratio;heldout_inlier_rmse;forward_reverse_translation;forward_reverse_rotation
-8,13,18.141079267476005,28.43334478424675,0.020491803278688523,0.1759614106055459,6,128.98300559552638,1.9608884223852157,1.569223825596656,0.0,False,backend_not_converged;heldout_inlier_ratio;heldout_inlier_rmse;forward_reverse_translation;forward_reverse_rotation;multistart_instability
-9,10,1.9264207261313253,56.00549858403757,0.6576312576312576,0.1069384415347781,6,15.95767235456931,1.6379095873833018,4.808769928108965,1.0,False,forward_reverse_translation;forward_reverse_rotation
-9,11,4.747620352849464,34.63405146967702,0.5883320678309288,0.11815838246331258,6,21.526672921842795,2.059598786777497,11.82652375960602,1.0,False,forward_reverse_translation;forward_reverse_rotation
-9,12,7.208566899019793,16.356227217122623,0.5413589364844904,0.12469816852190199,6,38.03547459177405,1.1412129496099401,4.794166723137469,1.0,False,forward_reverse_translation;forward_reverse_rotation
-9,13,10.706998136562337,16.406001445523756,0.015145729922362225,0.16400783300033744,6,309.6688821904962,3.999785287368469,12.657822343539058,0.5,False,heldout_inlier_ratio;heldout_inlier_rmse;forward_reverse_translation;forward_reverse_rotation
-9,14,7.911071381405571,14.085282580602351,0.5416463116756228,0.12859551377809345,6,26.721469573230685,0.011377143482079926,0.04588334980819398,1.0,True,
-10,11,3.0731501091601263,21.371447114360553,0.7131414267834794,0.12095106901516853,6,13.404202373771934,0.006897671932216771,0.18540056788520015,1.0,True,
-10,12,6.011368280631378,39.649271366914945,0.6117876278616659,0.1250022412120491,6,19.806559061723252,0.006509808900810373,0.0674238161099294,1.0,True,
-10,13,9.76425648692991,72.41150002956132,0.5244808055380743,0.1368926377190841,6,36.182352720806286,0.004651842966001154,0.17931102925270942,1.0,True,
-10,14,6.554187411792988,41.92021600343522,0.5996858385693572,0.12955691635611982,6,17.916357473627126,0.009593578345611885,0.0575378323241282,1.0,True,
-10,15,10.1706917218607,65.51298166784419,0.5048970366649924,0.13966622273060353,6,30.173625507677606,0.006074999657627903,0.048675594115713976,1.0,True,
-11,12,3.3258193587172027,18.277824252554396,0.6508076728924785,0.12017753597340275,6,15.211760062254436,0.016302173123952872,0.05141023051579196,1.0,True,
-11,13,7.195203402213438,51.04005291520078,0.5666710199817161,0.12966061077144855,6,26.1420797484443,0.010250004424021028,0.06307533415790957,1.0,True,
-11,14,3.634158560526847,20.548768889074672,0.64271407110666,0.12109534664165013,6,14.045896850674039,0.0076106764286992265,0.053501024445426364,1.0,True,
-11,15,7.446972831184529,44.14153455348363,0.570479416362689,0.12836105648415896,6,18.857231663145765,0.006898635427401595,0.028532809464236104,1.0,True,
-11,16,10.651790397923806,69.98495270981525,0.47336531178995206,0.13436934972977357,6,42.57049752296035,0.03569532774909474,0.4066994385898772,1.0,True,
-12,13,3.8697507070400543,32.762228662646386,0.7056733087955325,0.1168985924651875,6,14.509757354686162,0.0020816591736723326,0.08889045468170857,1.0,True,
-12,14,0.9871080784265185,2.2709446365202766,0.8745432399512789,0.09766070609034913,6,11.857755748379178,0.0024736851957500175,0.01806957610594206,1.0,True,
-12,15,4.171948395051693,25.86371030092923,0.7033426183844012,0.1210390118956012,6,11.859397045728281,0.02278023379659275,0.04927694758545221,1.0,True,
-12,16,7.331699780686257,51.70712845726085,0.5820235756385069,0.1245813395619955,6,32.351864267271324,0.009898398498331785,0.03808519306435069,1.0,True,
-12,17,6.344245780495681,1.4248238883471156,0.6542219994988725,0.12658997017247905,6,11.229930872030522,0.019742666743374927,0.07899239993213694,1.0,True,
-13,14,3.7992314627329202,30.491284026126113,0.6984766461034874,0.11406275275916469,6,13.924716870947337,0.012533601308866885,0.10861598809330086,1.0,True,
-13,15,0.9105848166450461,6.898518361717151,0.8794391298650243,0.0990320912382964,6,10.514819201987361,0.006638809913640662,0.04266354358349535,1.0,True,
-13,16,3.4957081323467,18.94489979461447,0.7273073505141552,0.10958532316684444,6,19.405056504086563,0.005441055528599314,0.12519365495729431,1.0,True,
-13,17,2.9366461487378266,31.337404774299262,0.7130265716137395,0.11779895987026272,6,12.816390530620648,0.013263327701592529,0.16221305155705523,1.0,True,
-13,18,5.248032013425982,3.445996452937746,0.7019876443728176,0.11940768524727288,6,25.985486690964827,0.008995185112868311,0.05198334816510605,1.0,True,
-14,15,3.8816793634199405,23.592765664408958,0.7089927153981411,0.11692319124843933,6,10.742828632896593,0.09073157715426228,0.8466374858545868,0.5,False,forward_reverse_translation;forward_reverse_rotation
-14,16,7.29101265002769,49.43618382074057,0.5923489278752436,0.12241125802320883,6,33.653658158107916,0.003104270324855819,0.13386021564092,1.0,True,
-14,17,5.915950814087913,0.8461207481731609,0.6687795177728063,0.12452036369781098,6,8.953051466023156,0.011592867275090568,0.10132037554601482,1.0,True,
-14,18,8.433581718971825,27.04528757318836,0.6196476790536196,0.12845389972151766,6,17.805501350543512,0.010071755472376367,0.13041952825792016,1.0,True,
-14,19,9.056542482242314,79.47127430600666,0.5845660749506904,0.12584132662356765,6,30.335272352492893,0.016830803029543952,0.25747616717579747,1.0,True,
-15,16,3.579287497246505,25.843418156331627,0.7032674772036475,0.1160221689285562,6,22.696890954829914,0.0009988867448377137,0.0903616813411077,1.0,True,
-15,17,2.2400625117908257,24.438886412582114,0.7489009568140678,0.11923375870790395,6,6.944925131742929,0.02582458487951321,0.09503968088936432,1.0,True,
-15,18,4.6956742726068,3.452521908779405,0.7165438713998661,0.1178392296206422,6,19.334165264362177,0.009354386824150452,0.1747840133793935,1.0,True,
-15,19,5.176369821469625,55.87850864159772,0.6924358974358974,0.11506895717743917,6,20.417126084614868,0.012497900854286311,0.3255615001296683,1.0,True,
-15,20,1.1877416357686716,128.54504202158432,0.6751867872591427,0.11901811643083532,6,27.381712286843683,2.670174543103617,1.9448445706434312,1.0,False,forward_reverse_translation;forward_reverse_rotation
-16,17,2.956793995513645,50.28230456891372,0.6284461152882206,0.11136186275509914,6,25.18428705575399,0.015372505619716304,0.0751987172187524,1.0,True,
-16,18,3.369822545690391,22.390896247552213,0.6921281286473868,0.10919235768364494,6,39.42922894015897,0.005355462743716019,0.036543030274398446,1.0,True,
-16,19,2.313038703742191,30.035090485266096,0.8880188913745961,0.09683468613705355,6,11.709094307553842,0.006745004702224827,0.04851926363284082,1.0,True,
-16,20,4.242518045976368,102.70162386525263,0.36698412698412697,0.13430062225326117,6,102.05441227788522,1.6194077849835204,3.5461435258834046,1.0,False,forward_reverse_translation;forward_reverse_rotation
-16,21,9.189957911290794,153.31148960700713,0.23764328854924197,0.14863424676851908,6,75.99451631707726,0.01721606559516444,0.561012451825117,0.5,False,heldout_inlier_ratio;forward_reverse_rotation
-17,18,2.517968959880004,27.89140832136152,0.7665916015366274,0.11462271017395576,6,9.759734204828526,0.003677259621955875,0.03455865038654393,1.0,True,
-17,19,3.518310045065406,80.31739505417983,0.645738203957382,0.11467441399973677,6,29.935027493013713,0.002657916871055147,0.03996637099866608,1.0,True,
-17,20,3.4199679241812992,152.98392843416642,0.2749902761571373,0.1378092692558536,6,36.98799912625142,0.9962499249666478,0.8583377913930432,0.5,False,heldout_inlier_ratio;forward_reverse_translation;forward_reverse_rotation
-17,21,8.339875108212771,156.4062058240796,0.4672368255565338,0.13751533768592306,6,32.19730139309397,3.7101311942981123,2.4525134629880094,0.0,False,forward_reverse_translation;forward_reverse_rotation;multistart_instability
-17,22,10.944066586184825,150.8296621811644,0.4255952380952381,0.14178517078123323,6,38.71138397815585,1.8838258929727458,2.643696557684808,0.0,False,forward_reverse_translation;forward_reverse_rotation;multistart_instability
-18,19,2.0416624616211574,52.42598673281832,0.6999343401181878,0.10876407223187459,6,36.91910298018587,0.0029722527816906435,0.028553700929610345,1.0,True,
-18,20,5.836864764777489,125.0925201128049,0.5796614723267061,0.12778176089815013,6,32.46060078153021,0.08496447343578362,0.24110896992060832,0.5,False,forward_reverse_translation
-18,21,10.84206419743439,175.70238585457497,0.23118979432439468,0.15492240575842026,6,95.57603256530417,0.3127879752995947,1.8359225701448998,1.0,False,heldout_inlier_ratio;forward_reverse_translation;forward_reverse_rotation
-18,22,13.461930516546937,178.72107050264088,0.20872354073123797,0.1539528469442395,6,158.96979855906298,5.509854679183967,5.400016950581102,1.0,False,heldout_inlier_ratio;forward_reverse_translation;forward_reverse_rotation
-18,23,10.787990019880349,164.1156966348418,0.3967277486910995,0.14073241205403234,6,67.81571155690442,0.037998014107336324,0.09667096210109852,1.0,True,
-19,20,6.120105723142265,72.6665333799865,0.6260444787247719,0.1266849163783949,6,25.979821491778758,0.02278172414929769,0.1983479736758361,1.0,True,
-19,21,11.201089261967727,123.27639912174082,0.22215292503430212,0.15090955425182226,6,96.8936157485445,2.289236102137041,0.9775110057344923,0.0,False,heldout_inlier_ratio;forward_reverse_translation;forward_reverse_rotation;multistart_instability
-19,22,13.962230939038237,128.85294276465584,0.1883148831488315,0.1538013519108862,6,137.2212791172918,1.512900854675742,1.7779489198947585,1.0,False,heldout_inlier_ratio;forward_reverse_translation;forward_reverse_rotation
-19,23,10.877481155608614,143.45831663234054,0.37768025078369905,0.1432975121556297,6,38.93505210462377,2.6484748594978824,0.9762972275431752,0.0,False,forward_reverse_translation;forward_reverse_rotation;multistart_instability
-19,24,10.384369483528566,177.87107120381796,0.43504761904761907,0.1392444656116938,6,41.9366060613255,0.04258350776081601,0.7879276702809289,1.0,False,forward_reverse_rotation
-20,21,5.091648376409225,50.60986574175432,0.6607188376242672,0.12214671257866083,6,14.082798146569486,0.014717307768564562,0.07480162341419787,1.0,True,
-20,22,7.842914217245693,56.186409384669375,0.576328684508104,0.13245799460807808,6,16.034011252152304,0.021651469263481868,0.4612054468064915,1.0,True,
-20,23,4.953146569484805,70.79178325235414,0.6451819579702717,0.1210226317867539,6,12.51146899612653,0.018249896901540018,0.12115759970964086,1.0,True,
-20,24,4.806112840058592,105.20453782384023,0.7383177570093458,0.11441422046802803,6,12.152737785424252,0.009011280440944078,0.06430292826078875,1.0,True,
-20,25,7.7432318481763245,68.57831319871164,0.6182822702159718,0.12791257682460658,6,29.104161441720713,0.01470353026057929,0.3205553574328468,1.0,True,
-21,22,2.836151129858731,5.576543642915048,0.7740636818348177,0.11564789267022943,6,12.209471497858695,0.006518431057241462,0.06991931948827856,1.0,True,
-21,23,1.4635995041292513,20.181917510599828,0.862223327530465,0.10307112004551743,6,11.124777032517395,0.002678668765812511,0.01879593375546071,1.0,True,
-21,24,2.7001854883863183,54.59467208208592,0.7795265676152102,0.11182158240581809,6,15.934013426145448,0.003491995501354943,0.037425651095358885,1.0,True,
-21,25,3.6513937480713023,17.968447456957325,0.7340892465252378,0.12002239056891176,6,16.572162980052227,0.03773466739435234,0.28781074838303833,1.0,True,
-21,26,4.368847767445087,60.912845043424156,0.7147358216190014,0.11997311573294335,6,17.3723156532691,0.01216183417643751,0.10312122071404906,1.0,True,
-22,23,3.806551883906871,14.605373867684776,0.7408951563458002,0.1172672510975272,6,12.610213323576234,0.009206244303296198,0.0960198600148152,1.0,True,
-22,24,4.999470711928798,49.01812843917085,0.6936064556176288,0.11914624513148228,6,13.20338761495324,0.006090737153725476,0.02755713841749006,1.0,True,
-22,25,2.281002791409386,12.391903814042275,0.7356584485868911,0.11474070552638106,6,16.673633938013026,0.001961709159757097,0.011643770804742994,1.0,True,
-22,26,1.990287035474152,55.33630140050909,0.7422594142259414,0.11765221626900067,6,18.880420396501982,0.007814937371704422,0.03934255356487579,1.0,True,
-22,27,2.5532406896142295,80.6206767037528,0.7254925373134329,0.11870181965154772,6,21.309389349405333,0.018241823604788293,0.08279236858581901,1.0,True,
-23,24,1.2663665558774873,34.41275457148608,0.7884810126582279,0.10662565692629541,6,10.611199681165658,0.0018823381482244372,0.020239211377623904,1.0,True,
-23,25,5.050865141821003,2.213470053642503,0.6843137254901961,0.1217109193489842,6,15.15080906413716,0.01463985673592735,0.20460048921133533,1.0,True,
-23,26,5.595299803053147,40.730927532824325,0.6772228989037758,0.12237954766026346,6,16.382334072569808,0.04518647006837217,0.20367965681897174,1.0,True,
-23,27,6.240759831138249,66.01530283606803,0.6637469586374696,0.12340951265232125,6,20.64383986204704,0.010626193556947943,1.1817079481882706,1.0,False,forward_reverse_rotation
-23,28,6.598772106458927,85.87759904182478,0.6720351390922401,0.12122778564083768,6,19.206570679040215,0.043540468662592216,0.2605147805386839,0.5,True,
-24,25,6.316196707646632,36.626224625128586,0.6764267990074442,0.12367636388755723,6,21.50176872604184,0.03581640576644142,0.20465043373191275,1.0,True,
-24,26,6.840027061556236,6.318172961338242,0.6524044389642417,0.12740222503880924,6,21.75476709306229,0.0620474433075452,0.12014838295938772,1.0,True,
-24,27,7.477875812552711,31.60254826458195,0.6546798029556651,0.12425657449629156,6,26.295019203914542,0.05403266260961897,0.2126374478532295,1.0,True,
-24,28,7.81303641449596,51.464844470338676,0.6614377470355731,0.12010155595807544,6,20.985000954928537,0.04949067679791638,0.25138526875322614,0.5,True,
-24,29,7.4949096209288655,93.73390958258972,0.047106325706594884,0.16491171897379944,6,28.651559198797667,0.8785910837751835,4.6775880176920355,0.0,False,heldout_inlier_ratio;heldout_inlier_rmse;forward_reverse_translation;forward_reverse_rotation;multistart_instability
-25,26,1.433673687807028,42.944397586466835,0.9137395459976105,0.08148335762110498,6,16.56285514245561,0.003236736725553251,0.006670817258604747,1.0,True,
-25,27,1.973107678535245,68.22877288971054,0.8596658711217183,0.10961458820309757,6,22.188153380460914,0.010624789877375612,0.04475651255675051,1.0,True,
-25,28,2.578633669986193,88.09106909546726,0.8839157491622786,0.10490699814192775,6,16.172179483480598,0.0028341913749953818,0.01585308444597317,1.0,True,
-25,29,1.4869736566208207,130.3601342077183,0.7857227558401518,0.10954358768244278,6,20.132087187715292,0.0032477187428175502,0.016542353808297643,1.0,True,
-25,30,5.843711828111692,139.6773015481418,0.05061061531235322,0.15827452276344428,6,186.52424080569762,2.4401001990983646,2.630476332375907,0.0,False,heldout_inlier_ratio;forward_reverse_translation;forward_reverse_rotation;multistart_instability
-26,27,0.6711664435459649,25.284375303243706,0.9188612099644128,0.07520973241900301,6,20.799051531446725,0.00099208733077853,0.005485007138530622,1.0,True,
-26,28,1.202247282991071,45.146671509000434,0.9182726623840114,0.07611768518866213,6,21.995419584098137,0.004984978187528897,0.011735070988964648,1.0,True,
-26,29,0.8313560685788559,87.41573662125148,0.7856890251090416,0.1085139306178217,6,24.280602674778585,0.002586732426994246,0.12205883610742861,1.0,True,
-26,30,5.554376083336843,177.37830086539105,0.7634835395750642,0.10495724850674515,6,34.80647378548566,0.008489213184549637,0.021733210949119494,1.0,True,
-26,31,6.5424104132673175,156.01119868987084,0.7374054682955207,0.10737521663044328,6,38.310555401782864,0.007807741115783734,0.04372865220115068,0.5,True,
-27,28,0.6147316450225401,19.862296205756735,0.9281131178707225,0.07220941715117642,6,20.092791877654474,0.002300778353404822,0.001589714984734129,1.0,True,
-27,29,0.7540837235696874,62.13136131800778,0.7871188037207112,0.10966659884725233,6,24.009650087098127,0.005134403698946511,0.024348440992038003,1.0,True,
-27,30,5.07252652550922,152.0939255621477,0.7922108208955224,0.10147035321534549,6,32.52026042532519,0.007488026390185518,0.03860522829819172,1.0,True,
-27,31,6.285022243874859,178.7044260068885,0.7592097617664149,0.10539218018667616,6,48.70669958381935,0.0020937297862771895,0.027401753528281184,1.0,True,
-27,32,5.778763736289045,138.47370648510574,0.7480278422273782,0.10408734715846861,6,42.21510199826032,0.004407463145301957,0.03490543724835993,0.5,True,
-28,29,1.3242039147826599,42.26906511225104,0.787814381863266,0.10901449493832827,6,25.9136254196751,0.01570175790237293,0.035696604981368125,1.0,True,
-28,30,5.091487440029177,132.23162935639098,0.7791159962581852,0.1043856075500982,6,36.50167792578953,0.011701496185342901,0.047047156803733815,1.0,True,
-28,31,6.538757407908195,158.84212980112872,0.7449015266285981,0.10668967013687278,6,48.04636971250768,0.02442137437725171,0.6810283060803413,1.0,False,forward_reverse_rotation
-28,32,5.963604730984572,158.33600269086236,0.7252210330386226,0.11213192172123396,6,59.78207706090093,0.01795092616246746,0.05811390678775167,0.0,False,multistart_instability
-28,33,5.80807626014008,67.05379893079936,0.6692465836255895,0.10871806386783625,6,54.67376255043743,0.132593515882677,0.6593919290562373,0.5,False,forward_reverse_translation;forward_reverse_rotation
-29,30,4.767831371266539,89.96256424413991,0.7320662880982732,0.10929719051930432,6,30.745355096911858,0.005463604154020641,0.01858348348785812,1.0,True,
-29,31,5.715598450796842,116.57306468887764,0.7254562254562255,0.11323336570178014,6,49.09678922599784,0.005077799585122041,0.07265894091769737,1.0,True,
-29,32,5.281749147864957,159.39493219688646,0.33356393404819557,0.1449978595558761,6,111.7227195582132,0.3588603464467423,0.15715476414253352,1.0,False,heldout_inlier_ratio;forward_reverse_translation
-29,33,4.779166013738703,109.3228640430504,0.2850467289719626,0.13990657628540273,6,178.4302458902832,0.39012209563155037,0.27153598748575447,0.0,False,backend_not_converged;heldout_inlier_ratio;forward_reverse_translation;multistart_instability
-30,31,2.604154101624297,26.610500444737717,0.8286237272623269,0.09646596945131831,6,41.03513305025278,0.018905314487389895,0.08632503464138006,1.0,True,
-30,32,1.69105635082049,69.43236795274656,0.8065326633165829,0.09815063400019147,6,35.25635856167186,0.007341509941520377,0.023973741904830165,1.0,True,
-30,33,3.2411277385703925,160.7145717128097,0.09253766757622493,0.15464713396746754,6,31.157195533215592,1.0417523955000538,7.5233256615684665,1.0,False,backend_not_converged;heldout_inlier_ratio;forward_reverse_translation;forward_reverse_rotation
-31,32,0.9137201114724802,42.82186750800884,0.8146841206602162,0.09914628416022428,6,34.3317318149268,0.0693687705829607,0.45703051311645604,1.0,True,
-31,33,1.4965271484681508,134.10407126807198,0.7551430598250177,0.1003246191461096,6,28.537349904719445,0.005559091155809675,0.033005318250111694,1.0,True,
-32,33,1.9169426499676907,91.28220376006315,0.7585270860380031,0.0987597723287551,6,32.396623704761396,2.8694889670610495,5.798276157700327,0.0,False,forward_reverse_translation;forward_reverse_rotation;multistart_instability
@@ -1,416 +0,0 @@
-{
-  "schema_version": 1,
-  "success": true,
-  "convention": "T_RTK_lidar maps raw LiDAR points into the RTK navigation frame",
-  "equation": "A_RTK_ij X = X B_LiDAR_ij",
-  "frames": {
-    "RTK": {
-      "origin": "GGA positioning reference point; confirm ANT1/reference antenna in receiver configuration",
-      "x_axis": "horizontal projection of the rawHeading baseline direction reported by the receiver",
-      "y_axis": "left",
-      "z_axis": "up",
-      "yaw_enu_deg": "90 - rawHeadingDeg"
-    },
-    "LiDAR": "raw LiDAR sensor frame"
-  },
-  "backend": "small_gicp",
-  "measured_lidar_extrinsic_used_as_initial": false,
-  "body_heading_offset_used": false,
-  "body_antenna_lever_xy_used": false,
-  "translation_m": [
-    1.6362863962126635,
-    -0.2461267175734333,
-    0.0851285837804841
-  ],
-  "rotation_rpy_deg_xyz": [
-    -0.7337275596198988,
-    1.3206501867905613,
-    -22.293138670917568
-  ],
-  "quaternion_xyzw": [
-    -0.004053851502643108,
-    0.012544688899275593,
-    -0.19323028574532955,
-    0.9810648570503344
-  ],
-  "matrix_4x4": [
-    [
-      0.9250093749023974,
-      0.37904117671318505,
-      0.026180960611867178,
-      1.6362863962126635
-    ],
-    [
-      -0.3792445939369631,
-      0.9252912459175456,
-      0.0031061548487011197,
-      -0.2461267175734333
-    ],
-    [
-      -0.023047653074967735,
-      -0.01280221013107426,
-      0.9996523941368298,
-      0.0851285837804841
-    ],
-    [
-      0.0,
-      0.0,
-      0.0,
-      1.0
-    ]
-  ],
-  "quality": {
-    "stations": 34,
-    "pairs": 55,
-    "residuals": {
-      "pairs": 55,
-      "translation_m": {
-        "rms": 0.17789650760700804,
-        "median": 0.07286630775287825,
-        "p90": 0.3720138200025907,
-        "p95": 0.3991391086271845,
-        "max": 0.5307044931342686
-      },
-      "rotation_deg": {
-        "rms": 1.6408745761259504,
-        "median": 0.7722797079864339,
-        "p90": 2.06911907015371,
-        "p95": 4.199030922414834,
-        "max": 4.936245135060849
-      },
-      "per_pair": [
-        {
-          "pair_index": 0,
-          "translation_m": 0.0467393979562504,
-          "rotation_deg": 0.7657431692138433
-        },
-        {
-          "pair_index": 1,
-          "translation_m": 0.05689017243056396,
-          "rotation_deg": 0.808177090866613
-        },
-        {
-          "pair_index": 2,
-          "translation_m": 0.06397107958463491,
-          "rotation_deg": 0.3801068497009622
-        },
-        {
-          "pair_index": 3,
-          "translation_m": 0.1107952950892596,
-          "rotation_deg": 0.22993541870282783
-        },
-        {
-          "pair_index": 4,
-          "translation_m": 0.025928748934271984,
-          "rotation_deg": 0.6046131588115704
-        },
-        {
-          "pair_index": 5,
-          "translation_m": 0.07562027851221909,
-          "rotation_deg": 0.6309936647228139
-        },
-        {
-          "pair_index": 6,
-          "translation_m": 0.18916050340904586,
-          "rotation_deg": 0.26477854089271946
-        },
-        {
-          "pair_index": 7,
-          "translation_m": 0.043365771415170416,
-          "rotation_deg": 0.10315887875456749
-        },
-        {
-          "pair_index": 8,
-          "translation_m": 0.06454458096453627,
-          "rotation_deg": 0.8075022372885624
-        },
-        {
-          "pair_index": 9,
-          "translation_m": 0.05200877185535076,
-          "rotation_deg": 0.5226003188288877
-        },
-        {
-          "pair_index": 10,
-          "translation_m": 0.08066933230596428,
-          "rotation_deg": 0.6774188851574047
-        },
-        {
-          "pair_index": 11,
-          "translation_m": 0.02679009789021938,
-          "rotation_deg": 0.5921424571930466
-        },
-        {
-          "pair_index": 12,
-          "translation_m": 0.033564310488394096,
-          "rotation_deg": 1.0085274567011704
-        },
-        {
-          "pair_index": 13,
-          "translation_m": 0.11139219509163403,
-          "rotation_deg": 0.596821136461582
-        },
-        {
-          "pair_index": 14,
-          "translation_m": 0.06985048714143455,
-          "rotation_deg": 0.8031764622169965
-        },
-        {
-          "pair_index": 15,
-          "translation_m": 0.05926582195242711,
-          "rotation_deg": 2.1146215228832355
-        },
-        {
-          "pair_index": 16,
-          "translation_m": 0.053822686491612724,
-          "rotation_deg": 0.5853967911734774
-        },
-        {
-          "pair_index": 17,
-          "translation_m": 0.06095128937580921,
-          "rotation_deg": 0.44465865069314414
-        },
-        {
-          "pair_index": 18,
-          "translation_m": 0.1465573557185661,
-          "rotation_deg": 2.0008653910594214
-        },
-        {
-          "pair_index": 19,
-          "translation_m": 0.10402727477620027,
-          "rotation_deg": 0.1402969578428506
-        },
-        {
-          "pair_index": 20,
-          "translation_m": 0.06790583241641802,
-          "rotation_deg": 0.31916987926311186
-        },
-        {
-          "pair_index": 21,
-          "translation_m": 0.037917260718891004,
-          "rotation_deg": 0.6351484564817018
-        },
-        {
-          "pair_index": 22,
-          "translation_m": 0.03171219066433979,
-          "rotation_deg": 0.7611907317100989
-        },
-        {
-          "pair_index": 23,
-          "translation_m": 0.08117084986905364,
-          "rotation_deg": 1.5477927872017732
-        },
-        {
-          "pair_index": 24,
-          "translation_m": 0.07160138896826257,
-          "rotation_deg": 0.47596695875013556
-        },
-        {
-          "pair_index": 25,
-          "translation_m": 0.01792823137987938,
-          "rotation_deg": 0.3736004835939083
-        },
-        {
-          "pair_index": 26,
-          "translation_m": 0.07286630775287825,
-          "rotation_deg": 1.800095203725142
-        },
-        {
-          "pair_index": 27,
-          "translation_m": 0.18109166588563658,
-          "rotation_deg": 3.9699177938767174
-        },
-        {
-          "pair_index": 28,
-          "translation_m": 0.24825286214410192,
-          "rotation_deg": 4.619204763488191
-        },
-        {
-          "pair_index": 29,
-          "translation_m": 0.06588464658827484,
-          "rotation_deg": 0.8156362703186536
-        },
-        {
-          "pair_index": 30,
-          "translation_m": 0.07851282552634996,
-          "rotation_deg": 4.018956419097684
-        },
-        {
-          "pair_index": 31,
-          "translation_m": 0.16064262686816105,
-          "rotation_deg": 4.8460088207546645
-        },
-        {
-          "pair_index": 32,
-          "translation_m": 0.1352330525899733,
-          "rotation_deg": 4.936245135060849
-        },
-        {
-          "pair_index": 33,
-          "translation_m": 0.0649121457474399,
-          "rotation_deg": 1.99675397761927
-        },
-        {
-          "pair_index": 34,
-          "translation_m": 0.0460266060768811,
-          "rotation_deg": 1.2713317047431354
-        },
-        {
-          "pair_index": 35,
-          "translation_m": 0.11269690515103371,
-          "rotation_deg": 0.5444667992697017
-        },
-        {
-          "pair_index": 36,
-          "translation_m": 0.05570004024342942,
-          "rotation_deg": 0.6200055232178192
-        },
-        {
-          "pair_index": 37,
-          "translation_m": 0.006720345812259607,
-          "rotation_deg": 1.1919322914717754
-        },
-        {
-          "pair_index": 38,
-          "translation_m": 0.05571822171452726,
-          "rotation_deg": 0.8100733260702963
-        },
-        {
-          "pair_index": 39,
-          "translation_m": 0.03156268585174415,
-          "rotation_deg": 1.1994412303651207
-        },
-        {
-          "pair_index": 40,
-          "translation_m": 0.1396566575693052,
-          "rotation_deg": 1.219665033277036
-        },
-        {
-          "pair_index": 41,
-          "translation_m": 0.42067987248571403,
-          "rotation_deg": 1.1901837024442368
-        },
-        {
-          "pair_index": 42,
-          "translation_m": 0.5307044931342686,
-          "rotation_deg": 1.3118186035265211
-        },
-        {
-          "pair_index": 43,
-          "translation_m": 0.0278193588287066,
-          "rotation_deg": 0.5463366974423033
-        },
-        {
-          "pair_index": 44,
-          "translation_m": 0.08083543060888142,
-          "rotation_deg": 1.597876847929754
-        },
-        {
-          "pair_index": 45,
-          "translation_m": 0.3628770836511988,
-          "rotation_deg": 0.7452934588576191
-        },
-        {
-          "pair_index": 46,
-          "translation_m": 0.40429181316256,
-          "rotation_deg": 1.3531101551442906
-        },
-        {
-          "pair_index": 47,
-          "translation_m": 0.3857777439966788,
-          "rotation_deg": 0.596976162782461
-        },
-        {
-          "pair_index": 48,
-          "translation_m": 0.18867103421064077,
-          "rotation_deg": 0.5750764247339077
-        },
-        {
-          "pair_index": 49,
-          "translation_m": 0.3781049775701853,
-          "rotation_deg": 1.067294104551723
-        },
-        {
-          "pair_index": 50,
-          "translation_m": 0.33164608192922734,
-          "rotation_deg": 0.4502803262827393
-        },
-        {
-          "pair_index": 51,
-          "translation_m": 0.39693080668345215,
-          "rotation_deg": 0.7722797079864339
-        },
-        {
-          "pair_index": 52,
-          "translation_m": 0.05933929949217937,
-          "rotation_deg": 0.9106743636918464
-        },
-        {
-          "pair_index": 53,
-          "translation_m": 0.09592936224298491,
-          "rotation_deg": 0.8770147335571411
-        },
-        {
-          "pair_index": 54,
-          "translation_m": 0.10907903367697035,
-          "rotation_deg": 0.527094410495204
-        }
-      ]
-    },
-    "weighted_jacobian_condition_number": 5.739204995062189,
-    "linearized_one_sigma": {
-      "translation_m": [
-        0.006273860396200483,
-        0.006271739194739563,
-        0.005766938056265489
-      ],
-      "rotation_deg": [
-        0.06648607826531162,
-        0.07036185324485841,
-        0.1461990497815168
-      ],
-      "warning": "conditional local estimate; bootstrap is the primary stability check"
-    },
-    "bootstrap": {
-      "runs": 100,
-      "order": [
-        "x_m",
-        "y_m",
-        "z_m",
-        "roll_deg",
-        "pitch_deg",
-        "yaw_deg"
-      ],
-      "std": [
-        0.002418391924231912,
-        0.0018782458573924745,
-        0.0024162629680468473,
-        0.09973116386354614,
-        0.0717005619911583,
-        0.09954619224880032
-      ],
-      "p025": [
-        1.6323740665381714,
-        -0.24963297512571514,
-        0.08157494287763181,
-        -0.9175278935375253,
-        1.1732724086153574,
-        -22.44062221365902
-      ],
-      "p975": [
-        1.6411098578257144,
-        -0.24215579438024784,
-        0.09016027720271011,
-        -0.5264292351234366,
-        1.4226725074325792,
-        -22.072092788653528
-      ]
-    }
-  },
-  "z_constraint": {
-    "observable_from_planar_AX_XB": false,
-    "method": "LiDAR ground planes plus externally supplied RTK reference-point height above ground",
-    "rtk_reference_height_above_ground_m": 0.8535,
-    "warning": "z is conditional on the supplied RTK antenna height; it is not independently identified by planar Ackermann motion"
-  },
-  "important_limit": "AX residual and bootstrap quantify internal consistency, not independent centimetre-grade absolute certification"
-}
@@ -1,48 +0,0 @@
-{
-  "final": {
-    "translation_m": [
-      1.6381793500373911,
-      -0.24084479868828831,
-      0.08448123595331278
-    ],
-    "rotation_rpy_deg_xyz": [
-      -0.8171674587248069,
-      1.323288118779805,
-      -22.104163317857477
-    ],
-    "pairs": 25,
-    "translation_rms_m": 0.10020667268070801,
-    "rotation_rms_deg": 1.2527941187072538,
-    "condition_number": 7.739413195936781
-  },
-  "backend_difference": {
-    "translation_m": 0.003889255293759414,
-    "rotation_deg": 0.1884307130161592,
-    "delta_matrix_4x4": [
-      [
-        0.9999968771376496,
-        0.0024968749848742764,
-        -0.00010644368722136346,
-        0.0008526003522697501
-      ],
-      [
-        -0.0024970968314049877,
-        0.9999945974960792,
-        -0.0021376356258303525,
-        -0.003658904827736509
-      ],
-      [
-        0.00010110570323801577,
-        0.0021378947504826257,
-        0.9999977095892133,
-        0.0010058801324768218
-      ],
-      [
-        0.0,
-        0.0,
-        0.0,
-        1.0
-      ]
-    ]
-  }
-}
@@ -0,0 +1,337 @@
+{
+  "schema_version": 1,
+  "success": true,
+  "convention": "T_RTK_lidar maps raw LiDAR points into the RTK navigation frame",
+  "equation": "A_RTK_ij X = X B_LiDAR_ij",
+  "frames": {
+    "RTK": {
+      "origin": "GGA positioning reference point; confirm ANT1/reference antenna in receiver configuration",
+      "x_axis": "vehicle forward after applying the configured G90 heading offset",
+      "y_axis": "left of the RTK X/baseline axis (not necessarily vehicle-left)",
+      "z_axis": "up",
+      "yaw_enu_deg": "90 - (rawHeadingDeg + -90)",
+      "frame_mode": "vehicle_forward_heading_offset"
+    },
+    "LiDAR": {
+      "description": "raw Helios sensor frame from points_raw polar decode",
+      "x_axis": "+X at azimuth 0° (forward when aviation connector faces vehicle rear)",
+      "y_axis": "+Y at azimuth +90° (left when +X is vehicle-forward)",
+      "z_axis": "up",
+      "origin_note": "optical/center per Helios manual; mounting height includes 63.5 mm base offset when deriving mechanical ΔZ"
+    }
+  },
+  "backend": "consensus",
+  "measured_lidar_extrinsic_used_as_initial": true,
+  "solver_initial_extrinsic": "D:\\First-dev-dept\\calibration-rtk-run\\run\\rtk_lidar_mechanical_initial.json",
+  "body_heading_offset_deg": -90.0,
+  "body_heading_offset_used": true,
+  "body_antenna_lever_xy_used": false,
+  "translation_m": [
+    0.21782224963960972,
+    -0.41134780227275347,
+    0.1065423366878719
+  ],
+  "rotation_rpy_deg_xyz": [
+    0.06623859235262805,
+    0.8096624730962496,
+    -0.5513220563826681
+  ],
+  "quaternion_xyzw": [
+    0.00061201333859398,
+    0.007062715262034178,
+    -0.004815137247499642,
+    0.9999632782988027
+  ],
+  "coordinate_contract_audit": {
+    "status": "no_near_180_degree_axis_conflict",
+    "requires_physical_axis_confirmation": false,
+    "mechanical_initial_path": "D:\\First-dev-dept\\calibration-rtk-run\\run\\rtk_lidar_mechanical_initial.json",
+    "mechanical_self_consistency": {
+      "baseline_points": "vehicle_right",
+      "frame_mode": "vehicle_forward_heading_offset",
+      "heading_offset_deg": -90.0,
+      "consistent": true,
+      "issues": []
+    },
+    "solution_vs_declared_baseline_side": {
+      "baseline_points": "vehicle_right",
+      "solution_yaw_deg": -0.5513220563826681,
+      "expected_yaw_deg": 0.0,
+      "yaw_error_deg": 0.5513220563826735,
+      "xy_error_m": 0.007516661287052368,
+      "z_error_m": 0.0280423356878719,
+      "mixed_translation_rotation_inheritance": false,
+      "near_expected_pose": true
+    },
+    "solution_relative_to_mechanical_initial": {
+      "translation_m": 0.029032271487700816,
+      "rotation_deg": 0.9820426535863694,
+      "delta_matrix_4x4": [
+        [
+          0.9998538650128304,
+          0.009638565806830942,
+          0.014119017957784103,
+          0.006962889639609726
+        ],
+        [
+          -0.009621275903042714,
+          0.9999528797859223,
+          -0.0012919976154994906,
+          0.002831671727246521
+        ],
+        [
+          -0.014130805670674627,
+          0.0011559658421926345,
+          0.9998994869856016,
+          0.0280423356878719
+        ],
+        [
+          0.0,
+          0.0,
+          0.0,
+          1.0
+        ]
+      ]
+    },
+    "note": "No automatic 180-degree correction was applied. Confirm static GNHPR left/right vs vehicle heading and Helios +X vs vehicle forward before deployment."
+  },
+  "matrix_4x4": [
+    [
+      0.9998538650128304,
+      0.009638565806830942,
+      0.014119017957784103,
+      0.21782224963960972
+    ],
+    [
+      -0.009621275903042714,
+      0.9999528797859223,
+      -0.0012919976154994906,
+      -0.41134780227275347
+    ],
+    [
+      -0.014130805670674627,
+      0.0011559658421926345,
+      0.9998994869856016,
+      0.1065423366878719
+    ],
+    [
+      0.0,
+      0.0,
+      0.0,
+      1.0
+    ]
+  ],
+  "quality": {
+    "stations": 27,
+    "pairs": 20,
+    "residuals": {
+      "pairs": 20,
+      "translation_m": {
+        "rms": 0.07116027693011169,
+        "median": 0.048258124379040895,
+        "p90": 0.10707349630410129,
+        "p95": 0.12100345665884242,
+        "max": 0.1859157912711514
+      },
+      "rotation_deg": {
+        "rms": 0.9820870524210346,
+        "median": 0.6015253089423158,
+        "p90": 1.626986719630708,
+        "p95": 1.7485896823567941,
+        "max": 2.6898815510329674
+      },
+      "per_pair": [
+        {
+          "pair_index": 0,
+          "translation_m": 0.10590532722733068,
+          "rotation_deg": 0.5565962122599133
+        },
+        {
+          "pair_index": 1,
+          "translation_m": 0.11758701799503664,
+          "rotation_deg": 0.36543717366192036
+        },
+        {
+          "pair_index": 2,
+          "translation_m": 0.07568813061789224,
+          "rotation_deg": 1.6990480050580474
+        },
+        {
+          "pair_index": 3,
+          "translation_m": 0.07918615709103373,
+          "rotation_deg": 0.33723851079591255
+        },
+        {
+          "pair_index": 4,
+          "translation_m": 0.04618587731787657,
+          "rotation_deg": 0.21044665580544528
+        },
+        {
+          "pair_index": 5,
+          "translation_m": 0.1859157912711514,
+          "rotation_deg": 2.6898815510329674
+        },
+        {
+          "pair_index": 6,
+          "translation_m": 0.015032558607699278,
+          "rotation_deg": 0.2283712127949818
+        },
+        {
+          "pair_index": 7,
+          "translation_m": 0.013237326674396453,
+          "rotation_deg": 1.6189799101387812
+        },
+        {
+          "pair_index": 8,
+          "translation_m": 0.07003654121179845,
+          "rotation_deg": 0.6464544056247182
+        },
+        {
+          "pair_index": 9,
+          "translation_m": 0.093450766078183,
+          "rotation_deg": 0.6476769159531274
+        },
+        {
+          "pair_index": 10,
+          "translation_m": 0.015054656507824072,
+          "rotation_deg": 1.041955302231996
+        },
+        {
+          "pair_index": 11,
+          "translation_m": 0.02338150242985638,
+          "rotation_deg": 0.9212619385214977
+        },
+        {
+          "pair_index": 12,
+          "translation_m": 0.01577916809950981,
+          "rotation_deg": 0.5021648028537498
+        },
+        {
+          "pair_index": 13,
+          "translation_m": 0.05033037144020521,
+          "rotation_deg": 0.31352985458971655
+        },
+        {
+          "pair_index": 14,
+          "translation_m": 0.059761151158357104,
+          "rotation_deg": 0.8078349560587313
+        },
+        {
+          "pair_index": 15,
+          "translation_m": 0.04160156756834453,
+          "rotation_deg": 0.32936437427192145
+        },
+        {
+          "pair_index": 16,
+          "translation_m": 0.02424387832291359,
+          "rotation_deg": 0.430799739443549
+        },
+        {
+          "pair_index": 17,
+          "translation_m": 0.05938006964566194,
+          "rotation_deg": 0.3855347569405268
+        },
+        {
+          "pair_index": 18,
+          "translation_m": 0.003922740066294881,
+          "rotation_deg": 0.8803664894801937
+        },
+        {
+          "pair_index": 19,
+          "translation_m": 0.02258390614293418,
+          "rotation_deg": 0.9489255319885953
+        }
+      ]
+    },
+    "weighted_jacobian_condition_number": 7.551077537197385,
+    "linearized_one_sigma": {
+      "translation_m": [
+        0.01109415838733884,
+        0.011043247019191056,
+        0.0049617743728656676
+      ],
+      "rotation_deg": [
+        0.08809558088476348,
+        0.08691570892756702,
+        0.13552099560800782
+      ],
+      "warning": "conditional local estimate; bootstrap is the primary stability check"
+    },
+    "bootstrap": {
+      "runs": 200,
+      "order": [
+        "x_m",
+        "y_m",
+        "z_m",
+        "roll_deg",
+        "pitch_deg",
+        "yaw_deg"
+      ],
+      "std": [
+        0.006235825930578342,
+        0.004851737742140203,
+        0.0004474159829207979,
+        0.07186455153709871,
+        0.11762318545143821,
+        0.061763211598034336
+      ],
+      "p025": [
+        0.20955504469154063,
+        -0.4174678620205485,
+        0.10568919463688603,
+        -0.06202286115194262,
+        0.5207464314096704,
+        -0.6639417037512721
+      ],
+      "p975": [
+        0.23211472114546824,
+        -0.4001091835786546,
+        0.1074353326322909,
+        0.23746760471402314,
+        1.0462704287205609,
+        -0.43455346485967655
+      ]
+    }
+  },
+  "z_constraint": {
+    "observable_from_planar_AX_XB": false,
+    "method": "LiDAR ground planes plus externally supplied RTK reference-point height above ground",
+    "rtk_reference_height_above_ground_m": 1.9165,
+    "warning": "z is conditional on the supplied RTK antenna height; it is not independently identified by planar Ackermann motion"
+  },
+  "important_limit": "AX residual and bootstrap quantify internal consistency, not independent centimetre-grade absolute certification",
+  "selection": {
+    "recommended": true,
+    "reason": "Uses only motion pairs accepted independently by both Open3D GICP and small_gicp",
+    "open3d_vs_small_gicp": {
+      "translation_m": 0.00312519750472982,
+      "rotation_deg": 0.11907006217018351,
+      "delta_matrix_4x4": [
+        [
+          0.9999996872332217,
+          -0.00030752729270551075,
+          -0.0007286703114418758,
+          0.000533303946557151
+        ],
+        [
+          0.00030892706731572284,
+          0.9999981058814015,
+          0.0019216653392719056,
+          -0.002909732881513971
+        ],
+        [
+          0.0007280779667146176,
+          -0.0019218898442212235,
+          0.9999978881187204,
+          0.001007919095162138
+        ],
+        [
+          0.0,
+          0.0,
+          0.0,
+          1.0
+        ]
+      ]
+    }
+  }
+}
@@ -0,0 +1,50 @@
+{
+  "final": {
+    "translation_m": [
+      0.21782224963960972,
+      -0.41134780227275347,
+      0.1065423366878719
+    ],
+    "rotation_rpy_deg_xyz": [
+      0.06623859235262805,
+      0.8096624730962496,
+      -0.5513220563826681
+    ],
+    "pairs": 20,
+    "translation_rms_m": 0.07116027693011169,
+    "rotation_rms_deg": 0.9820870524210346,
+    "condition_number": 7.551077537197385,
+    "coordinate_contract_status": "no_near_180_degree_axis_conflict",
+    "recommended_for_deployment": true
+  },
+  "backend_difference": {
+    "translation_m": 0.00312519750472982,
+    "rotation_deg": 0.11907006217018351,
+    "delta_matrix_4x4": [
+      [
+        0.9999996872332217,
+        -0.00030752729270551075,
+        -0.0007286703114418758,
+        0.000533303946557151
+      ],
+      [
+        0.00030892706731572284,
+        0.9999981058814015,
+        0.0019216653392719056,
+        -0.002909732881513971
+      ],
+      [
+        0.0007280779667146176,
+        -0.0019218898442212235,
+        0.9999978881187204,
+        0.001007919095162138
+      ],
+      [
+        0.0,
+        0.0,
+        0.0,
+        1.0
+      ]
+    ]
+  }
+}
@@ -6,18 +6,18 @@
 |---|---|
 | `run_full_pipeline.ps1` | 站目录导出 `combined/` 后跑到 `T_RTK_lidar` |
 | `export_multisensor_stations.ps1` | 薄封装：`tools/export_raw_to_combined.py` |
-| `prepare_multisensor_dataset.ps1` | 每站一帧 + yaw-only RTK 位姿 |
-| `run_direct_rtk_lidar.ps1` | 从 `combined/` 标定并封装最终结果（**默认基线系**） |
+| `prepare_multisensor_dataset.ps1` | 每站一帧 + RTK 位姿（默认含双天线 pitch/roll） |
+| `run_direct_rtk_lidar.ps1` | 从 `combined/` 标定并封装最终结果（**默认车头向前 -90**） |
 | `run_single_dataset.ps1` | 地面、双 GICP、精筛、共识、AX=XB |
 | `run_joint_rtk_lidar.ps1` | 多批共识对联合求解 |
 | `view_result.ps1` | 3D 运动对对比 |
 | `rtk_lidar_mechanical_initial.json` | 仅 AX=XB 初值；**禁止**用于 pair |

-## 默认参数（匹配当前约 2 m 车顶雷达 / 基线系）
+## 默认参数（匹配当前约 2 m 车顶雷达 / 车头向前）

 | 参数 | 默认 |
 |---|---|
-| `HeadingOffsetDeg` | `0`（基线系） |
+| `HeadingOffsetDeg` | `-90`（车头向前；主从装反、基线朝右） |
 | `GroundZMin/Max` | `-2.5` / `-1.5` |
 | `ExpectedStations` | `27` |
 | `MinStations` | `20` |
@@ -53,10 +53,10 @@ python tools\export_g90_h32_windows_to_combined.py `
 ```powershell
 powershell.exe -NoProfile -ExecutionPolicy Bypass -File "$Repo\run\run_direct_rtk_lidar.ps1" `
  -CombinedRoot "D:\data\rtk_lidar_run\combined" `
-  -WorkRoot "D:\data\rtk_lidar_run\prepared_baseline_h19165" `
-  -OutputRoot "D:\data\rtk_lidar_run\outputs_baseline_h19165" `
+  -WorkRoot "D:\data\rtk_lidar_run\prepared_vehicle_h19165" `
+  -OutputRoot "D:\data\rtk_lidar_run\outputs_vehicle_h19165" `
  -RtkReferenceHeightAboveGroundM 1.9165 `
-  -HeadingOffsetDeg 0 `
+  -HeadingOffsetDeg -90 `
  -ExpectedStations 27 `
  -GroundZMin -2.5 -GroundZMax -1.5
 ```
@@ -65,8 +65,8 @@ powershell.exe -NoProfile -ExecutionPolicy Bypass -File "$Repo\run\run_direct_rt

 ```powershell
 powershell.exe -NoProfile -ExecutionPolicy Bypass -File "$Repo\run\view_result.ps1" `
-  -Frames "D:\data\rtk_lidar_run\prepared_baseline_h19165\frames_all" `
-  -Pairs "D:\data\rtk_lidar_run\outputs_baseline_h19165\consensus\B_consensus.npz" `
-  -Extrinsic "D:\data\rtk_lidar_run\outputs_baseline_h19165\final_T_RTK_lidar.json" `
+  -Frames "D:\data\rtk_lidar_run\prepared_vehicle_h19165\frames_all" `
+  -Pairs "D:\data\rtk_lidar_run\outputs_vehicle_h19165\consensus\B_consensus.npz" `
+  -Extrinsic "D:\data\rtk_lidar_run\outputs_vehicle_h19165\final_T_RTK_lidar.json" `
  -PairIndex 0
 ```
@@ -1,23 +1,41 @@
 {
-  "schema_version": 1,
-  "convention": "T_RTK_lidar maps raw LiDAR points into the RTK baseline frame (X = rawHeading baseline, Y left, Z up; heading_offset_deg = 0)",
-  "translation_m": [
-    0.414179474,
+  "schema_version": 3,
+  "convention": "T_RTK_lidar maps raw LiDAR points into the vehicle-forward RTK body frame (X forward, Y left, Z up) after HeadingOffsetDeg=-90",
+  "frame_mode": "vehicle_forward_heading_offset",
+  "heading_offset_deg": -90.0,
+  "baseline_points": "vehicle_right",
+  "baseline_points_note": "Field-confirmed: master/slave assignment reversed vs G90 diagram, antennas left-right symmetric about rear-axle centerline. Master/GGA on vehicle left, slave on right; rawHeading points vehicle right.",
+  "vehicle_flu_lever_master_to_lidar_m": [
    0.210859360,
-    0.004000001
+    -0.414179474,
+    0.078500001
+  ],
+  "vehicle_flu_note": "Vehicle FLU: LiDAR origin relative to master/GGA = ahead, right, above. CAD drawing X was opposite vehicle-forward; longitudinal sign is +X in true FLU (solver also converges to +X).",
+  "antenna_symmetry_note": "Master/slave are mirrors about the rear-axle centerline; swap flips baseline 180° and the vehicle-Y sign of the master→LiDAR lever",
+  "translation_m": [
+    0.210859360,
+    -0.414179474,
+    0.078500001
  ],
  "rotation_rpy_deg_xyz": [
    0.0,
    0.0,
-    90.0
+    0.0
  ],
  "matrix_4x4": [
-    [0.0, -1.0, 0.0, 0.414179474],
    [1.0, 0.0, 0.0, 0.210859360],
-    [0.0, 0.0, 1.0, 0.004000001],
+    [0.0, 1.0, 0.0, -0.414179474],
+    [0.0, 0.0, 1.0, 0.078500001],
    [0.0, 0.0, 0.0, 1.0]
  ],
  "use": "Final AX=XB solver initialization only; never use for LiDAR pair registration",
-  "yaw_note": "≈90 deg yaw is expected when LiDAR X is vehicle-forward and the dual-antenna baseline is left-right",
-  "z_note": "CAD/mechanical z only; final z is constrained by measured GGA/ANT1 phase-center height above ground"
+  "yaw_note": "In vehicle-forward delivery, LiDAR +X ≈ vehicle forward ⇒ mechanical yaw ≈ 0",
+  "z_note": "78.500001 mm = H_L - H_R with H_L=1994.999879 mm, H_R=1916.499878 mm",
+  "attitude_composition": "R_W_body = Rz(yaw_raw) Ry(-pitch) Rx(roll) Rz(-heading_offset); pitch/roll stay in baseline frame",
+  "baseline_frame_equivalent": {
+    "heading_offset_deg": 0.0,
+    "translation_m": [0.414179474, 0.210859360, 0.078500001],
+    "rotation_rpy_deg_xyz": [0.0, 0.0, 90.0],
+    "note": "Same physical install expressed in rawHeading baseline frame"
+  }
 }
@@ -14,8 +14,9 @@ param(
    [int]$SmallGicpMaxGap = 26,
    [int]$Open3DMaxGap = 26,
    [double]$MaxReferenceTranslationM = 8.0,
-    # Baseline frame: rawHeading as RTK X. Use 90 only when deliberately targeting vehicle-forward.
-    [double]$HeadingOffsetDeg = 0.0,
+    # Baseline frame: rawHeading as RTK X. Default vehicle-forward for this car: -90
+    # (master/slave swapped, baseline points vehicle-right).
+    [double]$HeadingOffsetDeg = -90.0,
    [string]$SolverInitialExtrinsic = "",
    [double]$RefineMinInlierRatio = 0.63,
    [double]$RefineMaxInlierRmseM = 0.14
@@ -15,7 +15,7 @@
    # Roof-mounted H32 (~2 m). Do not reuse [-1.4, -0.4] on this vehicle.
    [double]$GroundZMin = -2.5,
    [double]$GroundZMax = -1.5,
-    [double]$HeadingOffsetDeg = 0.0
+    [double]$HeadingOffsetDeg = -90.0
 )

 $ErrorActionPreference = "Stop"
@@ -12,8 +12,12 @@ ROOT = Path(__file__).resolve().parents[1]
 sys.path.insert(0, str(ROOT / "tools"))
 sys.path.insert(0, str(ROOT / "code"))

-from finalize_direct_rtk_lidar import coordinate_contract_audit  # noqa: E402
+from finalize_direct_rtk_lidar import (  # noqa: E402
+    coordinate_contract_audit,
+    mechanical_self_consistency,
+)
 from prepare_multisensor_station_dataset import heading_to_enu_yaw  # noqa: E402
+from rtk_attitude import attitude_rotation, rtk_body_rotation  # noqa: E402
 from rigorous_calibration import (  # noqa: E402
    build_parser,
    load_extrinsic_matrix,
@@ -34,6 +38,32 @@ def test_east_vehicle_heading_maps_to_zero_enu_yaw() -> None:
    assert math.degrees(yaw) == 0.0


+def test_attitude_rotation_applies_baseline_pitch_elevation() -> None:
+    _, yaw = heading_to_enu_yaw(0.0, 0.0)  # heading north → body X = +North
+    rotation = attitude_rotation(yaw, pitch_deg=10.0, roll_deg=0.0)
+    body_x = rotation @ np.array([1.0, 0.0, 0.0])
+    np.testing.assert_allclose(
+        body_x,
+        [0.0, math.cos(math.radians(10.0)), math.sin(math.radians(10.0))],
+        atol=1e-12,
+    )
+
+
+def test_vehicle_forward_offset_keeps_pitch_about_baseline() -> None:
+    # Baseline points east (vehicle right if nose north); pitch elevates baseline X.
+    # Vehicle-forward offset -90 must not simply Ry after vehicle yaw.
+    raw_heading = 90.0
+    pitch = 10.0
+    r_correct = rtk_body_rotation(raw_heading, -90.0, pitch_deg=pitch, roll_deg=0.0)
+    _, yaw_raw = heading_to_enu_yaw(raw_heading, 0.0)
+    r_baseline = attitude_rotation(yaw_raw, pitch_deg=pitch, roll_deg=0.0)
+    rz90 = np.array([[0.0, -1.0, 0.0], [1.0, 0.0, 0.0], [0.0, 0.0, 1.0]])
+    np.testing.assert_allclose(r_correct, r_baseline @ rz90, atol=1e-12)
+    # Level vehicle-forward X should point north.
+    r_level = rtk_body_rotation(raw_heading, -90.0, pitch_deg=0.0, roll_deg=0.0)
+    np.testing.assert_allclose(r_level @ np.array([1.0, 0.0, 0.0]), [0.0, 1.0, 0.0], atol=1e-12)
+
+
 def test_pair_registration_has_no_extrinsic_argument() -> None:
    parser = build_parser()
    pair_options = {
@@ -45,18 +75,50 @@ def test_pair_registration_has_no_extrinsic_argument() -> None:
    assert "--global-voxel" in pair_options


-def test_mechanical_initial_round_trip() -> None:
+def test_mechanical_initial_is_vehicle_forward_swapped_master() -> None:
    path = ROOT / "run" / "rtk_lidar_mechanical_initial.json"
+    document = __import__("json").loads(path.read_text(encoding="utf-8-sig"))
    transform = load_extrinsic_matrix(path)
-    np.testing.assert_allclose(transform[:3, 3], [0.414179474, 0.210859360, 0.004000001])
-    np.testing.assert_allclose(transform[:3, :3], [[0.0, -1.0, 0.0], [1.0, 0.0, 0.0], [0.0, 0.0, 1.0]])
+    np.testing.assert_allclose(transform[:3, 3], [0.210859360, -0.414179474, 0.078500001])
+    np.testing.assert_allclose(transform[:3, :3], np.eye(3))
    np.testing.assert_allclose(params_transform(transform_params(transform)), transform, atol=1e-12)
+    assert document["baseline_points"] == "vehicle_right"
+    assert document["frame_mode"] == "vehicle_forward_heading_offset"
+    assert document["heading_offset_deg"] == -90.0
+    assert document["rotation_rpy_deg_xyz"][2] == 0.0
+    check = mechanical_self_consistency(document)
+    assert check["consistent"] is True
+
+
+def test_mixed_left_xy_plus_right_yaw_mechanical_is_rejected() -> None:
+    mixed = {
+        "baseline_points": "vehicle_left",
+        "translation_m": [0.414179474, 0.210859360, 0.078500001],
+        "rotation_rpy_deg_xyz": [0.0, 0.0, 90.0],
+        "matrix_4x4": [
+            [0.0, -1.0, 0.0, 0.414179474],
+            [1.0, 0.0, 0.0, 0.210859360],
+            [0.0, 0.0, 1.0, 0.078500001],
+            [0.0, 0.0, 0.0, 1.0],
+        ],
+    }
+    check = mechanical_self_consistency(mixed)
+    assert check["consistent"] is False
+
+
+def test_deprecated_minus_xy_right_baseline_is_rejected_for_swapped_master() -> None:
+    deprecated = {
+        "baseline_points": "vehicle_right",
+        "translation_m": [-0.414179474, -0.210859360, 0.078500001],
+        "rotation_rpy_deg_xyz": [0.0, 0.0, 90.0],
+    }
+    check = mechanical_self_consistency(deprecated)
+    assert check["consistent"] is False


 def test_near_180_degree_solution_is_flagged_for_physical_axis_check() -> None:
    initial_path = ROOT / "run" / "rtk_lidar_mechanical_initial.json"
    initial = load_extrinsic_matrix(initial_path)
-    # Flip the declared mechanical forward axis by ~180 deg about Z.
    solution = np.eye(4)
    solution[:3, :3] = initial[:3, :3] @ np.diag([-1.0, -1.0, 1.0])
    solution[:3, 3] = initial[:3, 3]
@@ -66,3 +128,26 @@ def test_near_180_degree_solution_is_flagged_for_physical_axis_check() -> None:
    })
    assert audit["status"] == "near_180_degree_axis_conflict"
    assert audit["requires_physical_axis_confirmation"] is True
+
+
+def test_previous_mixed_result_branch_is_not_recommended() -> None:
+    """Old baseline-frame mixed solution disagrees with vehicle-forward mechanical initial."""
+    initial_path = ROOT / "run" / "rtk_lidar_mechanical_initial.json"
+    solution = np.array(
+        [
+            [0.00942353438668686, -0.9999215926659111, 0.00824654595155475, 0.4123055815579212],
+            [0.9998714355322929, 0.009529416150714898, 0.012895837871912157, 0.2173092104098051],
+            [-0.012973411511822136, 0.008123961367132958, 0.9998828390593821, 0.10405760639434848],
+            [0.0, 0.0, 0.0, 1.0],
+        ],
+        float,
+    )
+    audit = coordinate_contract_audit({
+        "solver_initial_extrinsic": str(initial_path),
+        "matrix_4x4": solution.tolist(),
+    })
+    assert audit["requires_physical_axis_confirmation"] is True
+    assert audit["status"] in {
+        "near_180_degree_axis_conflict",
+        "solution_disagrees_with_mechanical_baseline_side",
+    }
@@ -183,7 +183,8 @@ def initialize_rtk_measurements(values: dict[str, np.ndarray]) -> None:
        ("differential_age_s", np.float64, np.nan),
        ("gnss_week", np.int32, -1), ("gnss_tow_ms", np.int64, -1),
        ("baseline_length_m", np.float64, np.nan), ("raw_heading_deg", np.float64, np.nan),
-        ("pitch_deg", np.float64, np.nan), ("heading_stddev_deg", np.float64, np.nan),
+        ("pitch_deg", np.float64, np.nan), ("roll_deg", np.float64, np.nan),
+        ("heading_stddev_deg", np.float64, np.nan),
        ("pitch_stddev_deg", np.float64, np.nan), ("heading_satellites", np.int32, -1),
        ("solution_satellites", np.int32, -1),
    ):
@@ -297,11 +298,15 @@ def build_combined(
                for key, dtype, default in (
                    ("gnss_week", np.int32, -1), ("gnss_tow_ms", np.int64, -1),
                    ("baseline_length_m", np.float64, np.nan), ("raw_heading_deg", np.float64, np.nan),
-                    ("pitch_deg", np.float64, np.nan), ("heading_stddev_deg", np.float64, np.nan),
+                    ("pitch_deg", np.float64, np.nan), ("roll_deg", np.float64, np.nan),
+                    ("heading_stddev_deg", np.float64, np.nan),
                    ("pitch_stddev_deg", np.float64, np.nan),
                    ("solution_satellites", np.int32, -1),
                ):
-                    values[f"rtk_{key}"] = np.asarray([heading_row.get(key, default)], dtype=dtype)
+                    value = heading_row.get(key, default)
+                    if key == "roll_deg" and value is None:
+                        value = 0.0
+                    values[f"rtk_{key}"] = np.asarray([value], dtype=dtype)
                values["rtk_heading_satellites"] = np.asarray([heading_row.get("satellites", -1)], dtype=np.int32)
                values["rtk_heading_solution_utf8"] = utf8_array(heading_row.get("heading_solution", ""))
                device_ns = gnss_utc_ns(heading_row, gps_utc_leap_seconds)
@@ -13,6 +13,8 @@ from typing import Any

 import numpy as np

+from rtk_attitude import heading_to_enu_yaw, rotation_to_quat_xyzw, rtk_body_rotation
+

 def args() -> argparse.Namespace:
    p = argparse.ArgumentParser(description=__doc__)
@@ -23,9 +25,48 @@ def args() -> argparse.Namespace:
    p.add_argument("--out", type=Path, required=True)
    p.add_argument("--max-bracket-ms", type=float, default=150.0)
    p.add_argument("--heading-std-limit-deg", type=float, default=0.5)
+    p.add_argument(
+        "--heading-offset-deg",
+        type=float,
+        default=None,
+        help="Added to rawHeading before ENU yaw. Default: body_heading_offset_deg from extrinsic JSON, else 0.",
+    )
+    p.add_argument(
+        "--orientation-model",
+        choices=("heading_pitch_roll", "yaw_only"),
+        default="heading_pitch_roll",
+        help="heading_pitch_roll uses GNHPR/UNIHEADINGA pitch+roll in T_W_RTK; yaw_only forces pitch=roll=0",
+    )
    return p.parse_args()


+POSITION_TYPES = {"GGA", "PVTSLNA"}
+HEADING_TYPES = {"UNIHEADINGA", "GNHPR"}
+
+
+def heading_row_valid(row: dict[str, Any]) -> bool:
+    if row.get("type") == "UNIHEADINGA":
+        return bool(row.get("checksum_valid") and row.get("heading_valid") and row.get("raw_heading_deg") is not None)
+    if row.get("type") == "GNHPR":
+        return bool(row.get("checksum_valid") and row.get("heading_valid") and row.get("raw_heading_deg") is not None)
+    return False
+
+
+def heading_quality_ok(row: dict[str, Any], std_limit_deg: float) -> list[str]:
+    reasons: list[str] = []
+    if row.get("type") == "UNIHEADINGA":
+        if str(row.get("heading_solution", "")) != "NARROW_INT":
+            reasons.append("HEADING_NOT_NARROW_INT")
+        std = float(row.get("heading_stddev_deg") or math.inf)
+        if std > std_limit_deg:
+            reasons.append("HEADING_STD_EXCEEDED")
+    elif row.get("type") == "GNHPR":
+        quality = int(row.get("heading_quality", -1) or -1)
+        if quality not in {4, 5} and not row.get("heading_valid"):
+            reasons.append("HEADING_QUALITY_NOT_FIXED")
+    return reasons
+
+
 def read_jsonl(path: Path) -> list[dict[str, Any]]:
    with path.open(encoding="utf-8") as f:
        return [json.loads(line) for line in f if line.strip()]
@@ -50,39 +91,6 @@ def ecef_to_enu(ecef: np.ndarray, origin: np.ndarray, lat_deg: float, lon_deg: f
    return r @ (ecef - origin)


-def yaw_matrix(yaw_rad: float) -> np.ndarray:
-    c, s = math.cos(yaw_rad), math.sin(yaw_rad)
-    return np.array([[c, -s, 0.0], [s, c, 0.0], [0.0, 0.0, 1.0]], dtype=float)
-
-
-def matrix_to_quat_xyzw(r: np.ndarray) -> np.ndarray:
-    # Stable branch-based conversion; output convention is x,y,z,w.
-    tr = float(np.trace(r))
-    if tr > 0.0:
-        s = math.sqrt(tr + 1.0) * 2.0
-        q = np.array([(r[2, 1] - r[1, 2]) / s,
-                      (r[0, 2] - r[2, 0]) / s,
-                      (r[1, 0] - r[0, 1]) / s, 0.25 * s])
-    else:
-        i = int(np.argmax(np.diag(r)))
-        if i == 0:
-            s = math.sqrt(1.0 + r[0, 0] - r[1, 1] - r[2, 2]) * 2.0
-            q = np.array([0.25 * s, (r[0, 1] + r[1, 0]) / s,
-                          (r[0, 2] + r[2, 0]) / s, (r[2, 1] - r[1, 2]) / s])
-        elif i == 1:
-            s = math.sqrt(1.0 + r[1, 1] - r[0, 0] - r[2, 2]) * 2.0
-            q = np.array([(r[0, 1] + r[1, 0]) / s, 0.25 * s,
-                          (r[1, 2] + r[2, 1]) / s, (r[0, 2] - r[2, 0]) / s])
-        else:
-            s = math.sqrt(1.0 + r[2, 2] - r[0, 0] - r[1, 1]) * 2.0
-            q = np.array([(r[0, 2] + r[2, 0]) / s,
-                          (r[1, 2] + r[2, 1]) / s, 0.25 * s,
-                          (r[1, 0] - r[0, 1]) / s])
-    if q[3] < 0.0:
-        q = -q
-    return q / np.linalg.norm(q)
-
-
 def bracket(rows: list[dict[str, Any]], times: np.ndarray, t: int,
            max_ns: int) -> tuple[dict[str, Any], dict[str, Any], float] | None:
    right = int(np.searchsorted(times, t, side="left"))
@@ -100,6 +108,10 @@ def circular_lerp_deg(a: float, b: float, u: float) -> float:
    return (a + u * delta) % 360.0


+def linear_lerp(a: float, b: float, u: float) -> float:
+    return (1.0 - u) * a + u * b
+
+
 def iso_utc(ns: int) -> str:
    return dt.datetime.fromtimestamp(ns / 1e9, dt.timezone.utc).isoformat(timespec="microseconds")

@@ -128,22 +140,38 @@ def main() -> int:
        lidar = [row for row in csv.DictReader(f) if not row.get("error")]
    rtk = read_jsonl(a.rtk_jsonl)
    imu = [row for row in read_jsonl(a.imu_jsonl) if row.get("crc_valid")]
-    gga = sorted([r for r in rtk if r.get("type") == "GGA" and r.get("checksum_valid")
-                  and r.get("lat_deg") is not None], key=lambda r: int(r["host_receive_utc_ns"]))
-    heading = sorted([r for r in rtk if r.get("type") == "UNIHEADINGA" and r.get("checksum_valid")
-                      and r.get("heading_valid") and r.get("raw_heading_deg") is not None],
-                     key=lambda r: int(r["host_receive_utc_ns"]))
-    if not lidar or len(gga) < 2 or len(heading) < 2:
-        raise RuntimeError("insufficient LiDAR/GGA/heading data")
+    positions = sorted(
+        [
+            r for r in rtk
+            if r.get("type") in POSITION_TYPES
+            and r.get("checksum_valid")
+            and r.get("lat_deg") is not None
+        ],
+        key=lambda r: int(r["host_receive_utc_ns"]),
+    )
+    heading = sorted(
+        [r for r in rtk if r.get("type") in HEADING_TYPES and heading_row_valid(r)],
+        key=lambda r: int(r["host_receive_utc_ns"]),
+    )
+    if not lidar or len(positions) < 2 or len(heading) < 2:
+        raise RuntimeError("insufficient LiDAR/GGA|PVTSLNA/heading(GNHPR|UNIHEADINGA) data")

    ext = json.loads(a.extrinsic.read_text(encoding="utf-8"))
    t_r_l = np.asarray(ext["matrix_4x4"], dtype=float)
    if t_r_l.shape != (4, 4):
        raise ValueError("extrinsic matrix_4x4 must be 4x4")
+    heading_offset_deg = (
+        float(a.heading_offset_deg)
+        if a.heading_offset_deg is not None
+        else float(ext.get("body_heading_offset_deg", 0.0) or 0.0)
+    )

-    gga_times = np.asarray([int(r["host_receive_utc_ns"]) for r in gga], dtype=np.int64)
+    position_times = np.asarray([int(r["host_receive_utc_ns"]) for r in positions], dtype=np.int64)
    heading_times = np.asarray([int(r["host_receive_utc_ns"]) for r in heading], dtype=np.int64)
-    origin_row = next(r for r in gga if int(r.get("fix_quality", -1)) == 4)
+    origin_row = next(
+        (r for r in positions if int(r.get("fix_quality", -1)) in {4, 5}),
+        positions[0],
+    )
    origin_lat, origin_lon, origin_alt = (float(origin_row[k]) for k in ("lat_deg", "lon_deg", "altitude_m"))
    origin_ecef = geodetic_to_ecef(origin_lat, origin_lon, origin_alt)
    max_ns = int(a.max_bracket_ms * 1_000_000)
@@ -151,7 +179,8 @@ def main() -> int:

    for index, frame in enumerate(lidar):
        t = int(frame["unix_time_ns"])
-        gb, hb = bracket(gga, gga_times, t, max_ns), bracket(heading, heading_times, t, max_ns)
+        gb = bracket(positions, position_times, t, max_ns)
+        hb = bracket(heading, heading_times, t, max_ns)
        reasons: list[str] = []
        available = gb is not None and hb is not None
        row: dict[str, Any] = {
@@ -160,7 +189,7 @@ def main() -> int:
            "pose_available": int(available), "gt_valid": 0, "invalid_reason": "",
        }
        if not available:
-            if gb is None: reasons.append("GGA_NOT_BRACKETED")
+            if gb is None: reasons.append("POSITION_NOT_BRACKETED")
            if hb is None: reasons.append("HEADING_NOT_BRACKETED")
            row.update({k: "" for k in ("x_m", "y_m", "z_m", "qx", "qy", "qz", "qw",
                                                 "rtk_x_m", "rtk_y_m", "rtk_z_m", "raw_heading_deg")})
@@ -174,31 +203,45 @@ def main() -> int:
        p1 = geodetic_to_ecef(float(g1["lat_deg"]), float(g1["lon_deg"]), float(g1["altitude_m"]))
        p_rtk = ecef_to_enu((1.0 - gu) * p0 + gu * p1, origin_ecef, origin_lat, origin_lon)
        raw_heading = circular_lerp_deg(float(h0["raw_heading_deg"]), float(h1["raw_heading_deg"]), hu)
-        yaw = math.radians(90.0 - raw_heading)
+        corrected_heading, yaw = heading_to_enu_yaw(raw_heading, heading_offset_deg)
+        if a.orientation_model == "heading_pitch_roll":
+            pitch = linear_lerp(float(h0.get("pitch_deg") or 0.0), float(h1.get("pitch_deg") or 0.0), hu)
+            roll = linear_lerp(float(h0.get("roll_deg") or 0.0), float(h1.get("roll_deg") or 0.0), hu)
+        else:
+            pitch = 0.0
+            roll = 0.0
        t_w_r = np.eye(4)
-        t_w_r[:3, :3] = yaw_matrix(yaw)
+        t_w_r[:3, :3] = rtk_body_rotation(
+            raw_heading, heading_offset_deg, pitch_deg=pitch, roll_deg=roll
+        )
        t_w_r[:3, 3] = p_rtk
        t_w_l = t_w_r @ t_r_l
-        q = matrix_to_quat_xyzw(t_w_l[:3, :3])
+        q = rotation_to_quat_xyzw(t_w_l[:3, :3])

        fix0, fix1 = int(g0.get("fix_quality", -1)), int(g1.get("fix_quality", -1))
-        sol0, sol1 = str(h0.get("heading_solution", "")), str(h1.get("heading_solution", ""))
-        std0 = float(h0.get("heading_stddev_deg") or math.inf)
-        std1 = float(h1.get("heading_stddev_deg") or math.inf)
-        if fix0 != 4 or fix1 != 4: reasons.append("RTK_POSITION_NOT_FIXED")
-        if sol0 != "NARROW_INT" or sol1 != "NARROW_INT": reasons.append("HEADING_NOT_NARROW_INT")
-        if max(std0, std1) > a.heading_std_limit_deg: reasons.append("HEADING_STD_EXCEEDED")
+        if fix0 not in {4, 5} or fix1 not in {4, 5}:
+            reasons.append("RTK_POSITION_NOT_FIXED")
+        reasons.extend(heading_quality_ok(h0, a.heading_std_limit_deg))
+        reasons.extend(heading_quality_ok(h1, a.heading_std_limit_deg))
+        # Deduplicate while preserving order
+        reasons = list(dict.fromkeys(reasons))
        row.update({
            "gt_valid": int(not reasons), "invalid_reason": ";".join(reasons),
            "x_m": t_w_l[0, 3], "y_m": t_w_l[1, 3], "z_m": t_w_l[2, 3],
            "qx": q[0], "qy": q[1], "qz": q[2], "qw": q[3],
            "rtk_x_m": p_rtk[0], "rtk_y_m": p_rtk[1], "rtk_z_m": p_rtk[2],
-            "raw_heading_deg": raw_heading, "yaw_enu_deg": math.degrees(yaw),
-            "gga_fix_before": fix0, "gga_fix_after": fix1,
-            "heading_solution_before": sol0, "heading_solution_after": sol1,
-            "heading_std_max_deg": max(std0, std1),
-            "gga_before_dt_ms": (t - int(g0["host_receive_utc_ns"])) / 1e6,
-            "gga_after_dt_ms": (int(g1["host_receive_utc_ns"]) - t) / 1e6,
+            "raw_heading_deg": raw_heading,
+            "corrected_heading_deg": corrected_heading,
+            "heading_offset_deg": heading_offset_deg,
+            "yaw_enu_deg": math.degrees(yaw),
+            "pitch_deg": pitch,
+            "roll_deg": roll,
+            "position_fix_before": fix0, "position_fix_after": fix1,
+            "heading_type_before": h0.get("type"), "heading_type_after": h1.get("type"),
+            "heading_solution_before": h0.get("heading_solution"),
+            "heading_solution_after": h1.get("heading_solution"),
+            "position_before_dt_ms": (t - int(g0["host_receive_utc_ns"])) / 1e6,
+            "position_after_dt_ms": (int(g1["host_receive_utc_ns"]) - t) / 1e6,
            "heading_before_dt_ms": (t - int(h0["host_receive_utc_ns"])) / 1e6,
            "heading_after_dt_ms": (int(h1["host_receive_utc_ns"]) - t) / 1e6,
        })
@@ -213,9 +256,19 @@ def main() -> int:

    summary = {
        "coordinate_convention": "T_W_L maps raw LiDAR points to local ENU; T_W_L = T_W_RTK @ T_RTK_lidar",
-        "world_frame": "local ENU, origin is the first RTK FIX GGA sample",
-        "rtk_frame": "x is rawHeading baseline direction projected horizontally, y left, z up",
-        "orientation_model": "RTK pose is yaw-only; IMU orientation is not fused",
+        "world_frame": "local ENU, origin is the first RTK FIX position sample",
+        "rtk_frame": (
+            "delivered body X follows rawHeading after heading_offset_deg; "
+            "pitch/roll applied in baseline frame before the fixed offset"
+        ),
+        "heading_offset_deg": heading_offset_deg,
+        "heading_sources_accepted": sorted(HEADING_TYPES),
+        "position_sources_accepted": sorted(POSITION_TYPES),
+        "orientation_model": a.orientation_model,
+        "orientation_composition": (
+            "R_W_body = Rz(yaw_raw) Ry(-pitch) Rx(roll) Rz(-heading_offset)"
+        ),
+        "orientation_note": "Uses dual-antenna GNHPR/UNIHEADINGA pitch/roll; IMU orientation is not fused",
        "time_basis": "LiDAR and serial host UTC; no jointly estimated clock offset/drift",
        "lidar_frames": len(pose_rows),
        "pose_available_frames": sum(int(r["pose_available"]) for r in pose_rows),
@@ -223,7 +276,10 @@ def main() -> int:
        "gt_invalid_frames": sum(not int(r["gt_valid"]) for r in pose_rows),
        "imu_frames": len(imu),
        "enu_origin": {"lat_deg": origin_lat, "lon_deg": origin_lon, "altitude_m": origin_alt},
-        "quality_rule": "GGA endpoints fix_quality=4, heading endpoints NARROW_INT, heading std <= limit, both streams bracket LiDAR time",
+        "quality_rule": (
+            "position endpoints fix_quality in {4,5}; UNIHEADINGA endpoints NARROW_INT with std gate; "
+            "GNHPR endpoints heading_valid/quality 4|5; both streams bracket LiDAR time"
+        ),
        "heading_std_limit_deg": a.heading_std_limit_deg,
        "max_bracket_ms": a.max_bracket_ms,
        "warning": "gt_valid is a quality gate, not independent proof of +/-3 cm absolute accuracy",
@@ -1,5 +1,5 @@
 #!/usr/bin/env python3
-"""Prepare one static LiDAR frame and one yaw-only RTK reference pose per NPZ segment."""
+"""Prepare one static LiDAR frame and one RTK reference pose per NPZ segment."""

 from __future__ import annotations

@@ -14,6 +14,13 @@ from typing import Any

 import numpy as np

+from rtk_attitude import (
+    heading_to_enu_yaw,
+    parse_pitch_roll_from_heading_raw,
+    rotation_to_quat_xyzw,
+    rtk_body_rotation,
+)
+
 POSE_FIELDS = ["time", "x", "y", "z", "qx", "qy", "qz", "qw"]


@@ -53,26 +60,30 @@ def ecef_to_enu(ecef: np.ndarray, origin: np.ndarray, lat_deg: float, lon_deg: f
    return rotation @ (ecef - origin)


-def yaw_rotation(yaw: float) -> np.ndarray:
-    c, s = math.cos(yaw), math.sin(yaw)
-    return np.array([[c, -s, 0.0], [s, c, 0.0], [0.0, 0.0, 1.0]])
-
-
-def heading_to_enu_yaw(raw_heading_deg: float, heading_offset_deg: float) -> tuple[float, float]:
-    """Convert GNHPR navigation heading to mathematical ENU yaw.
-
-    ``heading_offset_deg`` is added in the receiver's clockwise-from-north
-    heading convention. It is therefore not interchangeable with a ROS yaw
-    offset, whose sign and zero axis depend on the ROS frame definition.
-    """
-    corrected_heading = (raw_heading_deg + heading_offset_deg) % 360.0
-    return corrected_heading, math.radians(90.0 - corrected_heading)
-
-
-def scalar(data: np.lib.npyio.NpzFile, name: str) -> float:
+def scalar(data: np.lib.npyio.NpzFile, name: str, default: float | None = None) -> float:
+    if name not in data.files:
+        if default is None:
+            raise KeyError(name)
+        return float(default)
    return float(np.asarray(data[name]).reshape(-1)[0])


+def frame_pitch_roll(data: np.lib.npyio.NpzFile) -> tuple[float, float]:
+    pitch = scalar(data, "rtk_pitch_deg", math.nan)
+    roll = scalar(data, "rtk_roll_deg", math.nan)
+    if math.isfinite(pitch) and math.isfinite(roll):
+        return pitch, roll
+    raw = None
+    if "rtk_heading_raw_utf8" in data.files:
+        raw = bytes(np.asarray(data["rtk_heading_raw_utf8"]).reshape(-1))
+    parsed_pitch, parsed_roll = parse_pitch_roll_from_heading_raw(raw)
+    if not math.isfinite(pitch):
+        pitch = float(parsed_pitch) if parsed_pitch is not None else 0.0
+    if not math.isfinite(roll):
+        roll = float(parsed_roll) if parsed_roll is not None else 0.0
+    return pitch, roll
+
+
 def parse_args() -> argparse.Namespace:
    parser = argparse.ArgumentParser(description=__doc__)
    parser.add_argument("--combined-root", type=Path, required=True)
@@ -84,6 +95,12 @@ def parse_args() -> argparse.Namespace:
    parser.add_argument("--heading-std-limit-deg", type=float, default=0.5)
    parser.add_argument("--min-stations", type=int, default=30)
    parser.add_argument("--expected-stations", type=int, default=0)
+    parser.add_argument(
+        "--orientation-model",
+        choices=("heading_pitch_roll", "yaw_only"),
+        default="heading_pitch_roll",
+        help="heading_pitch_roll uses GNHPR/UNIHEADINGA pitch+roll; yaw_only forces roll=pitch=0",
+    )
    parser.add_argument("--overwrite", action="store_true")
    return parser.parse_args()

@@ -113,9 +130,10 @@ def main() -> int:
        for row in good:
            path = args.combined_root / Path(row["output"])
            with np.load(path, allow_pickle=False) as data:
+                pitch, roll = frame_pitch_roll(data)
                samples.append((scalar(data, "rtk_lat_deg"), scalar(data, "rtk_lon_deg"),
                                scalar(data, "rtk_altitude_m"), scalar(data, "rtk_raw_heading_deg"),
-                                scalar(data, "rtk_pitch_deg"), scalar(data, "rtk_heading_stddev_deg")))
+                                pitch, roll, scalar(data, "rtk_heading_stddev_deg", math.nan)))
        values = np.asarray(samples, dtype=float)
        heading_std = circular_std_deg(values[:, 3])
        if heading_std > args.heading_std_limit_deg:
@@ -123,16 +141,23 @@ def main() -> int:
            continue
        frame = good[len(good) // 2]
        source = args.combined_root / Path(frame["output"])
-        reported_std = values[:, 5]
+        reported_std = values[:, 6]
        reported_std_mean = float(np.nanmean(reported_std)) if np.isfinite(reported_std).any() else None
-        selected.append({"station": segment, "source": source, "time": int(frame["lidar_time_ns"]) / 1e9,
+        selected.append({
+            "station": segment, "source": source, "time": int(frame["lidar_time_ns"]) / 1e9,
            "lat": float(np.mean(values[:, 0])), "lon": float(np.mean(values[:, 1])),
-                         "alt": float(np.mean(values[:, 2])), "heading": circular_mean_deg(values[:, 3])})
-        summaries.append({"station": segment, "frames": len(group), "valid_fixed_frames": len(good),
+            "alt": float(np.mean(values[:, 2])), "heading": circular_mean_deg(values[:, 3]),
+            "pitch": float(np.mean(values[:, 4])), "roll": float(np.mean(values[:, 5])),
+        })
+        summaries.append({
+            "station": segment, "frames": len(group), "valid_fixed_frames": len(good),
            "heading_mean_deg": circular_mean_deg(values[:, 3]),
-                          "heading_circular_std_deg": heading_std, "rtk_pitch_mean_deg": float(np.mean(values[:, 4])),
+            "heading_circular_std_deg": heading_std,
+            "rtk_pitch_mean_deg": float(np.mean(values[:, 4])),
+            "rtk_roll_mean_deg": float(np.mean(values[:, 5])),
            "reported_heading_std_mean_deg": reported_std_mean,
-                          "altitude_std_m": float(np.std(values[:, 2])), "selected_source": str(source)})
+            "altitude_std_m": float(np.std(values[:, 2])), "selected_source": str(source),
+        })

    if args.expected_stations and len(selected) != args.expected_stations:
        raise RuntimeError(f"expected {args.expected_stations} usable stations, got {len(selected)}; rejected={rejected}")
@@ -145,6 +170,7 @@ def main() -> int:
    origin = selected[0]
    origin_ecef = geodetic_to_ecef(origin["lat"], origin["lon"], origin["alt"])
    lever = np.asarray(args.antenna_lever, dtype=float)
+    use_attitude = args.orientation_model == "heading_pitch_roll"
    pose_rows = []
    for index, item in enumerate(selected, 1):
        destination = frames / f"station_{index:02d}.npz"
@@ -152,32 +178,54 @@ def main() -> int:
        antenna = ecef_to_enu(geodetic_to_ecef(item["lat"], item["lon"], item["alt"]), origin_ecef,
                              origin["lat"], origin["lon"])
        corrected_heading, yaw = heading_to_enu_yaw(item["heading"], args.heading_offset_deg)
-        reference_position = antenna - yaw_rotation(yaw) @ lever
-        pose_rows.append(dict(zip(POSE_FIELDS, [item["time"], *reference_position, 0.0, 0.0,
-                                                math.sin(yaw / 2.0), math.cos(yaw / 2.0)])))
-        summaries[index - 1].update({"sequence": index, "prepared_frame": destination.name,
-                                     "corrected_heading_deg": corrected_heading})
+        pitch = float(item["pitch"]) if use_attitude else 0.0
+        roll = float(item["roll"]) if use_attitude else 0.0
+        rotation = rtk_body_rotation(
+            item["heading"], args.heading_offset_deg, pitch_deg=pitch, roll_deg=roll
+        )
+        reference_position = antenna - rotation @ lever
+        quat = rotation_to_quat_xyzw(rotation)
+        pose_rows.append(dict(zip(POSE_FIELDS, [item["time"], *reference_position, *quat])))
+        summaries[index - 1].update({
+            "sequence": index, "prepared_frame": destination.name,
+            "corrected_heading_deg": corrected_heading,
+            "pose_yaw_enu_deg": math.degrees(yaw),
+            "pose_pitch_deg": pitch, "pose_roll_deg": roll,
+        })
    pose_path = args.output / f"reference_poses_{args.pose_name}.csv"
    with pose_path.open("w", encoding="utf-8", newline="") as stream:
        writer = csv.DictWriter(stream, fieldnames=POSE_FIELDS); writer.writeheader(); writer.writerows(pose_rows)
    with (args.output / "station_summary.csv").open("w", encoding="utf-8", newline="") as stream:
        fields = sorted({key for row in summaries for key in row})
        writer = csv.DictWriter(stream, fieldnames=fields); writer.writeheader(); writer.writerows(summaries)
-    document = {"source_combined_root": str(args.combined_root.resolve()), "station_count": len(selected),
+    document = {
+        "source_combined_root": str(args.combined_root.resolve()), "station_count": len(selected),
        "rejected": rejected, "pose_csv": pose_path.name,
        "selection_policy": "middle LiDAR frame among fixed-position and valid-heading associations",
-                "reference_pose_configuration": {"raw_heading_offset_deg": args.heading_offset_deg,
+        "reference_pose_configuration": {
+            "raw_heading_offset_deg": args.heading_offset_deg,
            "antenna_lever_body_m": args.antenna_lever,
            "heading_offset_semantics": (
                "added to clockwise-from-north GNHPR heading before ENU yaw conversion"
            ),
-                    "orientation_model": "yaw-only, identical to the previous calibration workflow"},
+            "orientation_model": args.orientation_model,
+            "orientation_composition": (
+                "R_W_body = Rz(yaw_raw) Ry(-pitch) Rx(roll) Rz(-heading_offset); "
+                "yaw_raw from rawHeading, pitch/roll stay in baseline frame"
+            ),
+            "pitch_roll_note": (
+                "pitch/roll come from dual-antenna GNHPR/UNIHEADINGA (baseline elevation / reported roll). "
+                "This is not a fused IMU vehicle attitude; G90 roll is often ~0."
+            ),
+        },
        "stations": [{"sequence": i + 1, "source_station": item["station"],
                      "source_frame": str(item["source"]), "prepared_frame": f"station_{i + 1:02d}.npz"}
-                             for i, item in enumerate(selected)]}
+                     for i, item in enumerate(selected)],
+    }
    (args.output / "manifest.json").write_text(json.dumps(document, ensure_ascii=False, indent=2), encoding="utf-8")
    print(json.dumps({"prepared": str(args.output.resolve()), "stations": len(selected),
-                      "rejected": rejected, "pose_csv": pose_path.name}, ensure_ascii=False, indent=2))
+                      "rejected": rejected, "pose_csv": pose_path.name,
+                      "orientation_model": args.orientation_model}, ensure_ascii=False, indent=2))
    return 0


@@ -120,6 +120,7 @@ def parse_heading(line: str) -> dict:
        "baseline_length_m": safe_float(fields[2]),
        "raw_heading_deg": raw_heading,
        "pitch_deg": safe_float(fields[4]),
+        "roll_deg": 0.0,
        "heading_stddev_deg": safe_float(fields[6]),
        "pitch_stddev_deg": safe_float(fields[7]) if len(fields) > 7 else None,
        "station_id": fields[8].strip('"') if len(fields) > 8 else "",
@@ -0,0 +1,127 @@
+#!/usr/bin/env python3
+"""RTK dual-antenna attitude helpers shared by prepare and SLAM delivery."""
+
+from __future__ import annotations
+
+import math
+import re
+
+import numpy as np
+
+# GNHPR / UNIHEADINGA pitch is baseline elevation (far antenna higher ⇒ +pitch).
+# Build the baseline-frame attitude first, then apply the fixed body yaw offset:
+#   R_W_body = Rz(yaw_raw) Ry(-pitch) Rx(roll) Rz(-heading_offset)
+# so pitch/roll stay about the physical baseline, even when delivering vehicle-forward.
+
+
+def heading_to_enu_yaw(raw_heading_deg: float, heading_offset_deg: float = 0.0) -> tuple[float, float]:
+    """Convert clockwise-from-north heading to mathematical ENU yaw (rad)."""
+    corrected_heading = (raw_heading_deg + heading_offset_deg) % 360.0
+    return corrected_heading, math.radians(90.0 - corrected_heading)
+
+
+def _rz(yaw_rad: float) -> np.ndarray:
+    c, s = math.cos(yaw_rad), math.sin(yaw_rad)
+    return np.array([[c, -s, 0.0], [s, c, 0.0], [0.0, 0.0, 1.0]], dtype=float)
+
+
+def _ry(pitch_rad: float) -> np.ndarray:
+    c, s = math.cos(pitch_rad), math.sin(pitch_rad)
+    return np.array([[c, 0.0, s], [0.0, 1.0, 0.0], [-s, 0.0, c]], dtype=float)
+
+
+def _rx(roll_rad: float) -> np.ndarray:
+    c, s = math.cos(roll_rad), math.sin(roll_rad)
+    return np.array([[1.0, 0.0, 0.0], [0.0, c, -s], [0.0, s, c]], dtype=float)
+
+
+def attitude_rotation(
+    yaw_rad: float,
+    pitch_deg: float = 0.0,
+    roll_deg: float = 0.0,
+) -> np.ndarray:
+    """ENU←baseline rotation: Rz(yaw) Ry(-pitch) Rx(roll).
+
+    Positive ``pitch_deg`` elevates baseline X (slave higher than master).
+    """
+    return _rz(float(yaw_rad)) @ _ry(-math.radians(float(pitch_deg))) @ _rx(math.radians(float(roll_deg)))
+
+
+def rtk_body_rotation(
+    raw_heading_deg: float,
+    heading_offset_deg: float = 0.0,
+    pitch_deg: float = 0.0,
+    roll_deg: float = 0.0,
+) -> np.ndarray:
+    """ENU←delivered RTK body frame.
+
+    Pitch/roll are applied in the raw baseline frame; ``heading_offset_deg`` then
+    rotates that frame into the delivered body (0 = baseline X, -90 = vehicle
+    forward when baseline points vehicle-right on this vehicle).
+    """
+    _, yaw_baseline = heading_to_enu_yaw(raw_heading_deg, 0.0)
+    return attitude_rotation(yaw_baseline, pitch_deg, roll_deg) @ _rz(-math.radians(float(heading_offset_deg)))
+
+
+def rotation_to_quat_xyzw(rotation: np.ndarray) -> np.ndarray:
+    r = np.asarray(rotation, dtype=float)
+    tr = float(np.trace(r))
+    if tr > 0.0:
+        s = math.sqrt(tr + 1.0) * 2.0
+        q = np.array(
+            [(r[2, 1] - r[1, 2]) / s, (r[0, 2] - r[2, 0]) / s, (r[1, 0] - r[0, 1]) / s, 0.25 * s],
+            dtype=float,
+        )
+    else:
+        i = int(np.argmax(np.diag(r)))
+        if i == 0:
+            s = math.sqrt(1.0 + r[0, 0] - r[1, 1] - r[2, 2]) * 2.0
+            q = np.array(
+                [0.25 * s, (r[0, 1] + r[1, 0]) / s, (r[0, 2] + r[2, 0]) / s, (r[2, 1] - r[1, 2]) / s],
+                dtype=float,
+            )
+        elif i == 1:
+            s = math.sqrt(1.0 + r[1, 1] - r[0, 0] - r[2, 2]) * 2.0
+            q = np.array(
+                [(r[0, 1] + r[1, 0]) / s, 0.25 * s, (r[1, 2] + r[2, 1]) / s, (r[0, 2] - r[2, 0]) / s],
+                dtype=float,
+            )
+        else:
+            s = math.sqrt(1.0 + r[2, 2] - r[0, 0] - r[1, 1]) * 2.0
+            q = np.array(
+                [(r[0, 2] + r[2, 0]) / s, (r[1, 2] + r[2, 1]) / s, 0.25 * s, (r[1, 0] - r[0, 1]) / s],
+                dtype=float,
+            )
+    if q[3] < 0.0:
+        q = -q
+    return q / np.linalg.norm(q)
+
+
+def parse_pitch_roll_from_heading_raw(raw_utf8: bytes | str | None) -> tuple[float | None, float | None]:
+    """Best-effort pitch/roll from a stored GNHPR/UNIHEADINGA raw line."""
+    if raw_utf8 is None:
+        return None, None
+    text = raw_utf8.decode("ascii", "ignore") if isinstance(raw_utf8, (bytes, bytearray)) else str(raw_utf8)
+    text = text.strip()
+    if "GNHPR" in text:
+        parts = text.split(",")
+        if len(parts) >= 5:
+            try:
+                return float(parts[3]), float(parts[4])
+            except ValueError:
+                return None, None
+    if "UNIHEADINGA" in text.upper() or "HEADINGA" in text.upper():
+        payload = text.split(";", 1)[-1]
+        fields = payload.split(",")
+        if len(fields) >= 5:
+            try:
+                return float(fields[4]), 0.0
+            except ValueError:
+                return None, None
+    match = re.search(r",(-?\d+(?:\.\d+)?),(-?\d+(?:\.\d+)?),\d,", text)
+    if match:
+        try:
+            return float(match.group(1)), float(match.group(2))
+        except ValueError:
+            return None, None
+    return None, None
@@ -1,7 +1,7 @@
 # 雷达与 RTK 标定说明书

 本文说明如何用本仓库完成 **双天线 RTK ↔ 3D 激光雷达** 外参标定，得到可直接使用的 `T_RTK_lidar`。  
-默认交付坐标系为 **基线系**（`HeadingOffsetDeg = 0`）。更完整的指标与本次结果见根目录 [`README.md`](README.md)。
+默认交付坐标系为 **车头向前**（`HeadingOffsetDeg = -90`）。更完整的指标与本次结果见根目录 [`README.md`](README.md)。

 ---

@@ -14,8 +14,8 @@ p_RTK = T_RTK_lidar · p_lidar
 | 项目 | 说明 |
 |---|---|
 | 输出文件 | `final_T_RTK_lidar.json` |
-| 坐标系 | **基线系**：GGA 原点 + `rawHeading` 基线方向为 X（不是车体后轮轴系） |
-| 不用到的量 | 车体航向偏置、天线 XY 杆臂、IMU 姿态 |
+| 坐标系 | **车头向前**：GGA 原点 + 车头 X（本车 `HeadingOffsetDeg=-90`） |
+| 不用到的量 | 车体航向偏置、天线 XY 杆臂、IMU 融合姿态（双天线 pitch/roll 默认进入参考位姿） |
 | 必须提供 | RTK 参考点（通常 ANT1）**相位中心**离地高度 |
 | pair 配准 | **禁止**使用外参 seed |
 | 求解初值 | 可用 `run/rtk_lidar_mechanical_initial.json`（仅 AX=XB） |
@@ -92,7 +92,7 @@ powershell.exe -NoProfile -ExecutionPolicy Bypass -File "$Repo\run\run_full_pipe
  -ImuCapture "$Raw\captures\imu.rscap" `
  -OutputRoot $Out `
  -RtkReferenceHeightAboveGroundM 1.9165 `
-  -HeadingOffsetDeg 0 `
+  -HeadingOffsetDeg -90 `
  -ExpectedStations 27 `
  -GroundZMin -2.5 `
  -GroundZMax -1.5
@@ -101,7 +101,7 @@ powershell.exe -NoProfile -ExecutionPolicy Bypass -File "$Repo\run\run_full_pipe
 | 参数 | 含义 |
 |---|---|
 | `-RtkReferenceHeightAboveGroundM` | 相位中心离地高（m），**必填**；本车 1.9165 |
-| `-HeadingOffsetDeg` | 默认 0 = 基线系 |
+| `-HeadingOffsetDeg` | 默认 **-90** = 车头向前（本车主从装反） |
 | `-GroundZMin/Max` | 约 2 m 雷达用 `[-2.5,-1.5]` |
 | `-ExpectedStations` / `-MinStations` | 本批 27 / 20 |

@@ -110,16 +110,16 @@ powershell.exe -NoProfile -ExecutionPolicy Bypass -File "$Repo\run\run_full_pipe
 ```powershell
 powershell.exe -NoProfile -ExecutionPolicy Bypass -File "$Repo\run\run_direct_rtk_lidar.ps1" `
  -CombinedRoot "D:\data\rtk_lidar_run\combined" `
-  -WorkRoot "D:\data\rtk_lidar_run\prepared_baseline_h19165" `
-  -OutputRoot "D:\data\rtk_lidar_run\outputs_baseline_h19165" `
+  -WorkRoot "D:\data\rtk_lidar_run\prepared_vehicle_h19165" `
+  -OutputRoot "D:\data\rtk_lidar_run\outputs_vehicle_h19165" `
  -RtkReferenceHeightAboveGroundM 1.9165 `
-  -HeadingOffsetDeg 0 `
+  -HeadingOffsetDeg -90 `
  -ExpectedStations 27 `
  -GroundZMin -2.5 `
  -GroundZMax -1.5
 ```

-> 历史文档中的 **0.758 m / 34 站** 属于另一台车（data4），不要用于本车。
+> 不得复用其他车辆或历史采集的天线离地高度、站点数量与外参结果。

 ---

@@ -141,8 +141,8 @@ $OutputRoot/

 ```powershell
 $Repo = "D:\First-dev-dept\calibration-rtk-run"
-$Out  = "D:\data\rtk_lidar_run\outputs_baseline_h19165"
-$Work = "D:\data\rtk_lidar_run\prepared_baseline_h19165"
+$Out  = "D:\data\rtk_lidar_run\outputs_vehicle_h19165"
+$Work = "D:\data\rtk_lidar_run\prepared_vehicle_h19165"

 powershell.exe -NoProfile -ExecutionPolicy Bypass -File "$Repo\run\view_result.ps1" `
  -Frames "$Work\frames_all" `
Author	SHA1	Message	Date
lichun.qu	3c8e4f92f6	清理历史雷达RTK资料并归档当前车辆结果	2026-08-25 14:45:31 +08:00
lichun.quandCursor	5f59bcd795	改为车头向前整链：主从装反机械初值、双天线 pitch/roll 姿态与默认 HeadingOffsetDeg=-90 Co-authored-by: Cursor <cursoragent@cursor.com>	2026-08-11 18:07:54 +08:00
lichun.quandCursor	6242fd1081	删除 README 中远程旧一键复现说明小节 Co-authored-by: Cursor <cursoragent@cursor.com>	2026-08-10 22:58:46 +08:00