SLAM 點雲圖鑑Point clouds for construction
M-01 · 2024

Global BIM-point registration and association

作者把 BIM 構件以構造實體幾何（CSG）拆解並以解析距離場表示，避免取樣造成資訊損失。粗配準以平面基元對 BIM 面在重力軸對齊下搜尋對應，並以剛體動力學模擬驗證幾何一致性；精配準則交替更新位姿與逐點對應權重，並以鄰近性、法向與構件存在與否截斷權重，使臨時材料與未施作構件不誤導配準。模擬採 ISPRS 室內建模基準（含手持與背包掃描）。

下游工程任務近十年已讀全文同儕審查已出版營建應用研究

閱讀版本
version of record, ScienceDirect HTML (Automation in Construction 168 Part A, 105796, 1 December 2024), read in full
全文取得途徑
國立臺灣大學機構訂閱
本頁內容

Registers as-is point clouds to BIM globally using distance-field BIM primitives and jointly refines pose and BIM-point association with existence-aware weights to support progress monitoring.

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real site: handheld sensor suite with Ouster OS0-128 LiDAR (clouds built with FAST-LIO2)、simulation: ISPRS indoor modelling benchmark clouds from stationary, handheld and backpack scanners
原文測試平台handheld
狀態估計primitive-level coarse registration (orientation hypotheses using gravity, correspondence tree, geometric constraint filter, rigid-body dynamics simulator) + point-level Gauss-Newton fine registration with Geman-McClure weights (Sec. 3.2-3.3)
資料關聯plane primitives to BIM distance fields; per-point weights truncated by proximity, normal consistency and element existence (Sec. 3.3.2)
時間表示不適用
去畸變不適用
迴圈閉合不適用
全域最佳化不適用
地圖表示BIM as CSG-decomposed analytic distance fields; input point cloud
先驗資訊BIM (IFC) split into IfcWall, IfcSlab, IfcColumn, IfcBeam (and IfcCovering on site) and decomposed into convex CSG primitives with analytic distance fields; LOD 200-300 in simulation, LOD 300 on site; gravity axis assumed known in both frames, reducing 24 orientation hypotheses to 4 (Sec. 3.1, 3.2.1, 4.1.1, 4.2.1)
可輸出幾何BIM-aligned point cloud with per-point association weights (progress existence check)
計算需求Intel Core i9-12900H CPU; offline. Simulation: coarse registration 64.2 s on average (PLADE 8.85 s, RANSAC 27.27 s), fine registration 16.57 s (Sec. 4.1.2, Table 2). Real site: coarse about 2 min 45 s per floor, fine under 50 s (pose refinement 18.08 s plus association 30.66 s) (Sec. 4.2.2, Table 4); verification uses multi-threading and the Tsit5 ODE solver

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
LiDAROuster OS0-128方法輸入未標示on a handheld sensor suite; operator walked each floor; clouds reconstructed per floor with FAST-LIO2(Zhang et al., 2024b, Sec. 4.2.1)
地面雷射掃描儀（TLS）原文未報告 (stationary laser scanner)資料集感測器ISPRS benchmark on indoor modelling原文未報告(Zhang et al., 2024b, Sec. 4.1.1)
行動掃描設備原文未報告 (handheld and backpack laser scanners)資料集感測器ISPRS benchmark on indoor modelling原文未報告(Zhang et al., 2024b, Sec. 4.1.1)
運算硬體Intel(R) Core(TM) i9-12900H CPU執行運算平台未標示all experiments(Zhang et al., 2024b, Sec. 4)
其他原文未報告 (engineers' on-site measurement of 3-4 structural landmarks as GCPs)參考或真值量測未標示GCPs at different heights about 20 m apart; used with CloudCompare v2.13 alpha to build the reference registration(Zhang et al., 2024b, Sec. 4.2.1)

作者報告的優勢與限制

優勢

  • coarse registration median TE 0.053 m and RE 0.272 deg over the successful cases of 250 perturbed samples (Table 1)
  • fine registration median TE 0.0246 m vs 0.0422 m for point-to-point ICP (Table 2)
  • 100% coarse success for alpha_r >= 0.5 deg and alpha_t >= 0.4 m, unlike all ten baselines (Sec. 4.1.2, Fig. 10)
  • on seven floors of an active site, fine-registration errors of 0.009-0.203 deg and 0.032-0.107 m relative to the CloudCompare reference, about 0.05 deg and 0.058 m on average (Table 4)
  • per-point association separates rebar, glass, barriers, temporary materials, pipes and boxes from built structures and flags unbuilt decoration walls (Sec. 4.2.2, Fig. 15)

限制

  • drift errors in the FAST-LIO2 reconstruction caused some wall points to receive low association levels (floors 07 and 12); authors suggest a high-precision laser scanner (Sec. 4.2.2)
  • geometry-only association: barriers close to walls can be associated with the wall even with normal verification (Sec. 4.2.2)
  • only regular human-made structures are modelled; MEP objects are not included (Sec. 4.2.2)
  • offline only; point clouds with few planar segments cannot be registered (Sec. 4.2.2)
  • coarse registration slower than PLADE and RANSAC in simulation (64.2 s vs 8.85 s and 27.27 s) (Sec. 4.1.2)
  • coarse precision drops for sparse or partial clouds from early construction stages (Table 3)
  • (inference) the real-site reference is itself a manual CloudCompare registration seeded by 3-4 GCPs, so centimetre-level differences are not independently verified
  • (inference) assumes gravity axis known and planar primitives dominant

營建工程相關證據

實測於香港城市大學賽馬會一健康大樓施工工地 06 至 12 樓（CR Construction 協助，每層約 80 m × 50 m），以手持感測套件上的 Ouster OS0-128 蒐集資料，再以 FAST-LIO2 逐層重建點雲，BIM 為 LOD 300。評估用的參考值不是直接量測的點位誤差，而是以工程師現地量測的 3 至 4 個結構特徵點（GCP）作為同名點，在 CloudCompare v2.13 alpha 人工粗配準後再做點對網格精配準所得的轉換；屬施工中工地且使用 SLAM 點雲。

原文驗證環境：公開基準、施工中工地、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 6 個比較組，合計 88 筆紀錄。以下列出本方法紀錄最多的 4 組，其餘 2 組列在最後，並連到性能比較頁。

Zhang et al., 2024b · Table 4 本方法 42 筆

表格設定（擷取紀錄原文）：Jockey Club One Health Tower construction site, floors 06-12; handheld Ouster OS0-128 clouds reconstructed per floor with FAST-LIO2 and registered to a LOD 300 BIM without knowing the floor number; errors relative to a manual CloudCompare registration seeded by 3-4 engineer-measured GCP pairs and refined point-to-mesh (Zhang et al., 2024b, Table 4)

delta_r, rotation error，Jockey Club One Health Tower site data (self-collected) · Floor 06

只並列這張表在相同設定下報告的方法；以「本方法：」開頭者為本頁方法。失敗、未執行與未報告以標記呈現，不是 0。

這些是 Zhang et al., 2024b 在此表設定下報告的數值（author-reported results），只能在同一個比較組內對照，不代表方法在其他資料或設定下的表現。

統計量：原文未報告；對齊方式：未對齊；單位：deg；場景：active high-rise construction site, building interiors (Hong Kong)

資料來源作者報告值(Zhang et al., 2024b, Table 4)

數值與出處
方法（原文寫法）報告值出處
Ours (coarse registration)本方法原文提出0.039 deg(Zhang et al., 2024b, Table 4)
Ours (fine registration)本方法原文提出0.012 deg(Zhang et al., 2024b, Table 4)

Zhang et al., 2024b · Table 3 本方法 28 筆

表格設定（擷取紀錄原文）：Sensitivity of the proposed method: point clouds voxel-downsampled at different sizes, and partial clouds simulating temporal construction stages (Cases 01-03 from Models 01 and 02); mean errors (Zhang et al., 2024b, Table 3)

RE-Mean, rotation error，ISPRS benchmark on indoor modelling · Raw

只並列這張表在相同設定下報告的方法；以「本方法：」開頭者為本頁方法。失敗、未執行與未報告以標記呈現，不是 0。

這些是 Zhang et al., 2024b 在此表設定下報告的數值（author-reported results），只能在同一個比較組內對照，不代表方法在其他資料或設定下的表現。

統計量：平均值（mean）；對齊方式：未對齊；單位：deg；場景：building interiors from the ISPRS indoor modelling benchmark (Models 01-05)

資料來源作者報告值(Zhang et al., 2024b, Table 3)

數值與出處
方法（原文寫法）報告值出處
Ours (coarse registration)本方法原文提出0.125 deg(Zhang et al., 2024b, Table 3)
Ours (fine registration)本方法原文提出0.06 deg(Zhang et al., 2024b, Table 3)

Zhang et al., 2024b · Table 1 本方法 6 筆

資料集與序列ISPRS benchmark on indoor modelling · Models 01-05 (250 samples)

表格設定（擷取紀錄原文）：Coarse registration on 250 samples (50 per model) with random rigid perturbations (roll and pitch within 30 deg, yaw within 180 deg, translation within 10 m); only successful results with RE < 45 deg and TE < 10 m are included; A50, A75 and A95 quantiles of rotation and translation error against the benchmark alignment (Zhang et al., 2024b, Table 1)

RE_50, rotation error 50th percentile，ISPRS benchmark on indoor modelling · Models 01-05 (250 samples)

只並列這張表在相同設定下報告的方法；以「本方法：」開頭者為本頁方法。失敗、未執行與未報告以標記呈現，不是 0。

這些是 Zhang et al., 2024b 在此表設定下報告的數值（author-reported results），只能在同一個比較組內對照，不代表方法在其他資料或設定下的表現。

統計量：中位數（median）；對齊方式：未對齊；單位：deg；場景：building interiors from the ISPRS indoor modelling benchmark (Models 01-05)

資料來源作者報告值(Zhang et al., 2024b, Table 1)

數值與出處
方法（原文寫法）報告值出處
GMMTree (initialised with FPFH-RANSAC)4.466 deg(Zhang et al., 2024b, Table 1)
FilterReg (initialised with FPFH-RANSAC)3.59 deg(Zhang et al., 2024b, Table 1)
GO-ICP4.178 deg(Zhang et al., 2024b, Table 1)
Super4PCS2.208 deg(Zhang et al., 2024b, Table 1)
FGR27.139 deg(Zhang et al., 2024b, Table 1)
RANSAC (FPFH features)5.477 deg(Zhang et al., 2024b, Table 1)
RMMG1.673 deg(Zhang et al., 2024b, Table 1)
PLADE0.424 deg(Zhang et al., 2024b, Table 1)
DCP34.865 deg(Zhang et al., 2024b, Table 1)
PointNetLK11.741 deg(Zhang et al., 2024b, Table 1)
Ours (primitive-level coarse registration)本方法原文提出0.272 deg(Zhang et al., 2024b, Table 1)

Zhang et al., 2024b · Table 2 本方法 6 筆

資料集與序列ISPRS benchmark on indoor modelling · Models 01-05

表格設定（擷取紀錄原文）：Fine registration after coarse alignment on the simulation samples; errors against the benchmark alignment (Zhang et al., 2024b, Table 2)

Time [s], computation time per registration，ISPRS benchmark on indoor modelling · Models 01-05

只並列這張表在相同設定下報告的方法；以「本方法：」開頭者為本頁方法。失敗、未執行與未報告以標記呈現，不是 0。

這些是 Zhang et al., 2024b 在此表設定下報告的數值（author-reported results），只能在同一個比較組內對照，不代表方法在其他資料或設定下的表現。

統計量：平均值（mean）；對齊方式：未對齊；單位：s；場景：building interiors from the ISPRS indoor modelling benchmark (Models 01-05)

資料來源作者報告值(Zhang et al., 2024b, Table 2)

數值與出處
方法（原文寫法）報告值出處
ICP (point-to-point)硬體：Intel Core i9-12900H CPU16.34 s(Zhang et al., 2024b, Table 2)
ICP (point-to-plane)硬體：Intel Core i9-12900H CPU12.12 s(Zhang et al., 2024b, Table 2)
Ours (point-level fine registration with BIM-point association)本方法原文提出硬體：Intel Core i9-12900H CPU16.57 s(Zhang et al., 2024b, Table 2)

其他比較組

列出其餘 2 個比較組

來源

  • Zhang et al., 2024b

    Yinqiang Zhang, Liang Lu, Xiaowei Luo, Jia Pan（2024）Global BIM-point cloud registration and association for construction progress monitoringAutomation in Construction, 168 (Part A), 105796

    DOI 10.1016/j.autcon.2024.105796

    同儕審查已出版已讀全文近十年查證後修正

回到方法圖鑑

SLAM 點雲圖鑑

本站為綜述論文《SLAM-Based Point Clouds for Construction: A Review》研究過程的互動說明文件。互動圖為教學模擬，不代表任何特定系統的實測表現；方法描述、數值與結論以所引原始文獻為準。

內容基準日 2026-09-25 · 研究與網站製作使用 AI 輔助工具，見 證據標記與方法

選擇開啟Esc關閉