Kinematic LiDAR error budget
作者把 LiDAR 直接地理定位方程式中的 14 個觀測量（GNSS 位置、IMU 姿態、視準角、掃描角與距離、槓桿臂）做一階展開，以 Jacobian 將典型誤差傳遞為點位的水平與垂直精度，並模擬定翼機、直升機與地面車載三種平台。定翼機的水平誤差主要來自 IMU 與視準角誤差，作者歸納其水平誤差至少約為垂直誤差的 5 倍；直升機（Q-240）以光束發散造成的掃描角誤差為主，水平與垂直誤差比約為 2 至 2.5；地面系統的比值約為 2，誤差預算以雷射掃描儀本身為主，姿態誤差貢獻低於 25%。模型再以定翼機地面控制點、直升機連結點閉合差與既有地面系統高程檢核比對，預測值與實測相近。GNSS 誤差未納入模型，需另行相加。
本頁內容
First-order propagation of 14 georeferencing error sources to per-point horizontal and vertical accuracy for fixed-wing, helicopter and ground-vehicle LiDAR; attitude (IMU plus boresight) errors dominate fixed-wing horizontal error, scanner angle error from beam divergence dominates the helicopter case with a Riegl Q-240, and scanner errors dominate ground systems; predictions agree with control-point, tie-point and ground-control checks.
技術屬性
欄位內容為文獻擷取紀錄的原文用語（英文），以原文為據；「未查證」表示本研究尚未讀到該資訊，不代表該方法不具備此能力。
| 感測輸入 | LiDAR、GNSS、IMU |
|---|---|
| 原文測試平台 | fixed-wing aircraft (Terrapoint ALTMS, 1000 m AGL, 16 flight lines over an airport target site)、helicopter (Riegl Q-140 with Honeywell HG1700 IMU, 100 m AGL, four directions)、ground-based vehicle system with Honeywell HG1700 IMU (results reproduced from Glennie et al. 2006) |
| 狀態估計 | first-order (Taylor) error propagation of the direct georeferencing equation with 14 observed parameters (GNSS position, IMU roll, pitch and yaw, three boresight angles, scan angle, range, three lever-arm components) through the Jacobians J, K, B and C |
| 資料關聯 | 不適用 |
| 時間表示 | 不適用 |
| 去畸變 | 不適用 |
| 迴圈閉合 | 不適用 |
| 全域最佳化 | none |
| 地圖表示 | 不適用 |
| 先驗資訊 | post-processed DGPS/INS trajectory; assumed 1-sigma error magnitudes from specifications and experience (IMU attitude per Table 1, boresight per Table 2, 2 cm range, one quarter of beam divergence for footprint uncertainty, 2 cm lever arm); GNSS positioning error excluded from the model and added separately (2 cm) |
| 可輸出幾何 | expected horizontal and vertical accuracy |
| 計算需求 | 不適用 |
使用設備
原文使用的感測器、運算硬體與載具（equipment）。型號保留原文寫法，連結到設備頁中同一型號的歸併名稱；角色依原文用途分為方法輸入、資料集感測器、執行運算平台、參考或真值量測（reference or ground truth）與比較對象設備。
| 類別 | 型號（原文寫法） | 角色 | 資料集 | 原文規格 | 出處 |
|---|---|---|---|---|---|
| LiDAR | Terrapoint ALTMS | 方法輸入 | 未標示 | 510-class IMU, 0.75 mrad beam divergence, 2 cm ranging error, optimization-based boresight; flown at 1000 m AGL | (Glennie, 2007, Comparison section: fixed wing system; Table 6) |
| LiDAR | Riegl Q-140 | 方法輸入 | 未標示 | helicopter system with Honeywell HG1700 IMU; 100 m AGL, four directions, 115 tie points | (Glennie, 2007, Comparison section: helicopter system; Table 7) |
| LiDAR | Optech 3100 | 比較對象設備 | 未標示 | range error 0.02 m, angular resolution 0.001 deg, beam divergence 0.3 mrad, total angular error 0.0044 deg; used for fixed-wing simulations | (Glennie, 2007, Table 3) |
| LiDAR | Riegl Q-240 | 比較對象設備 | 未標示 | range error 0.02 m, angular resolution 0.005 deg, beam divergence 2.7 mrad, total angular error 0.039 deg; used for helicopter simulations | (Glennie, 2007, Table 3) |
| LiDAR | Riegl Q-280 | 比較對象設備 | 未標示 | range error 0.02 m, angular resolution 0.0025 deg, beam divergence 0.5 mrad, total angular error 0.0076 deg; used for helicopter simulations | (Glennie, 2007, Table 3) |
| 慣性量測單元（IMU） | Honeywell HG1700 | 方法輸入 | 未標示 | used in the helicopter validation system and the ground-based system; as Novatel SPAN (HG1700 AG58): 0.015 deg roll and pitch, 0.05 deg heading (Table 1) | (Glennie, 2007, Table 1; Comparison section) |
| 慣性量測單元（IMU） | Applanix 510 | 比較對象設備 | 未標示 | 0.005 deg roll and pitch, 0.008 deg heading (typical post-processed) | (Glennie, 2007, Table 1) |
| 慣性量測單元（IMU） | Applanix 610 | 比較對象設備 | 未標示 | 0.0025 deg roll and pitch, 0.005 deg heading (typical post-processed) | (Glennie, 2007, Table 1) |
| GNSS 接收器 | GPS base station (model not reported) | 方法輸入 | 未標示 | less than 1 km from the test area; 2 cm GPS error added to model predictions | (Glennie, 2007, Comparison section) |
| 其他 | ground targets (8 reflective horizontal and 8 vertical-only) | 參考或真值量測 | 未標示 | established at an airport calibration site; horizontal positions digitized from 1 m intensity rasters | (Glennie, 2007, Comparison section: fixed wing system) |
作者報告的優勢與限制
優勢
- Fixed-wing ALTMS validation: final RMSE 0.618 m horizontal and 0.086 m vertical versus model 0.553 m and 0.105 m (Table 6)
- Helicopter tie-point misclosure 0.292 m horizontal and 0.089 m vertical versus model 0.250 m and 0.095 m (Table 7)
- Ground system: model predicts 6 to 7 cm vertical including GPS error, observed RMS 0.059 m (Table 8 and text)
- Shows which error sources dominate per platform, e.g. attitude errors give 60% to 75% of fixed-wing horizontal error (Fig. 4 text)
限制
- Assumes GNSS/INS-based georeferencing; SLAM-derived trajectories without GNSS need a different trajectory error model (inference)
- GNSS positioning errors are excluded from the model and must be added by the user (Positioning errors section)
- Neglects incidence angle and terrain slope; beam footprint uncertainty fixed at one quarter of beam divergence (Laser scanner errors section)
- Models single-return accuracy only, not DEM or surface accuracy (Conclusions)
- Needs detailed a priori system information, which limits comparison with published studies (Comparison section)
營建工程相關證據
原文未報告
原文驗證環境：獨立參考量測
報告的性能數據
以下是原文作者報告的性能數值（author-reported results），不是本研究重新量測的結果。每張圖只並列同一個比較組（comparison group，同一張表、同一組實驗設定）內的方法；不同比較組之間的數值不可直接比較，也不構成排名。
本方法共出現在 3 個比較組，合計 5 筆紀錄。
Glennie, 2007 · Table 6 本方法 2 筆
指標Expected errors (model)
表格設定（擷取紀錄原文）：ALTMS fixed-wing LiDAR at 1000 m AGL, 16 flight lines over 16 airport targets; vertical from TIN of ground returns, horizontal from digitized 1 m intensity raster; Final RMSE accounts for half-pixel digitization error; model expectation includes 2 cm GPS error (Glennie, 2007, Table 6)
Expected errors (model)，Terrapoint production test, early 2006 · Horizontal
這張表在此指標與資料序列只列出本方法一筆，沒有可並列的其他方法，因此不畫圖，數值與出處見下表。這是 Glennie, 2007 在此表設定下報告的數值（author-reported results），不代表方法在其他資料或設定下的表現。
| 方法（原文寫法） | 報告值 | 出處 |
|---|---|---|
| 1st-order error model (expected)本方法原文提出 | 0.553 m | (Glennie, 2007, Table 6) |
Glennie, 2007 · Table 7 本方法 2 筆
指標Expected error (1st-order model)
表格設定（擷取紀錄原文）：Helicopter system (Riegl Q-140, Honeywell HG1700) at 100 m AGL over a calibration site flown in four directions; RMS misclosure of 115 tie points after least-squares boresighting versus model expectation (Glennie, 2007, Table 7)
Expected error (1st-order model)，Terrapoint helicopter boresight adjustment · Horizontal
這張表在此指標與資料序列只列出本方法一筆，沒有可並列的其他方法，因此不畫圖，數值與出處見下表。這是 Glennie, 2007 在此表設定下報告的數值（author-reported results），不代表方法在其他資料或設定下的表現。
| 方法（原文寫法） | 報告值 | 出處 |
|---|---|---|
| 1st-order error model (expected)本方法原文提出 | 0.25 m | (Glennie, 2007, Table 7) |
Glennie, 2007 · Text Comparison: Ground based system 本方法 1 筆
指標expected vertical accuracy from the error analysis
資料集與序列Glennie et al. 2006 ground system test · Vertical
表格設定（擷取紀錄原文）：Model prediction for the ground system of Table 8 (Glennie, 2007, Text Comparison: Ground based system)
expected vertical accuracy from the error analysis，Glennie et al. 2006 ground system test · Vertical
這張表在此指標與資料序列只列出本方法一筆，沒有可並列的其他方法，因此不畫圖，數值與出處見下表。這是 Glennie, 2007 在此表設定下報告的數值（author-reported results），不代表方法在其他資料或設定下的表現。
| 方法（原文寫法） | 報告值 | 出處 |
|---|---|---|
| 1st-order error model (expected)本方法原文提出 | 無數值未報告註記（擷取紀錄）：4 to 5 cm without GPS error; 6 to 7 cm after adding 2 cm GPS error | (Glennie, 2007, Comparison section: ground based system) |
來源
Glennie, 2007
（2007）Rigorous 3D error analysis of kinematic scanning LIDAR systemsJournal of Applied Geodesy, 1(3), pages 原文未報告 in Crossref
同儕審查已出版已讀全文經典