Year: 2025
Type: article
Abstract: LiDAR-based Simultaneous Localization and Mapping (SLAM) plays a key role in enabling inspection robots to achieve autonomous navigation. However, at installation construction sites of large-scale public buildings, existing methods often suffer from point-cloud drift, large z-axis errors, and inefficient loop closure detection, limiting their robustness and adaptability in complex environments. To address these issues, this paper proposes an improved algorithm, LeGO-LOAM-LPB (Large-scale Public Building), built upon the LeGO-LOAM framework. The method enhances feature quality through point-cloud preprocessing, stabilizes z-axis pose estimation by introducing ground-residual constraints, improves matching efficiency with an incremental k-d tree, and strengthens map consistency via a two-layer loop closure detection mechanism. Experiments conducted on a self-developed inspection robot platform in both simulated and real construction sites of large-scale public buildings demonstrate that LeGO-LOAM-LPB significantly improves positioning accuracy, reducing the root mean square error by 41.55% compared with the original algorithm. The results indicate that the proposed method offers a more precise and robust SLAM solution for safety inspection robots in construction environments and shows strong potential for engineering applications.
Source: Buildings
Authors: Chunyong Feng, Junqi Yu, Jingdan Li, Yonghua Wu, Ben Wang, Kaiwen Wang
Institution: Xi'an University of Architecture and Technology
Language: English
Subfield: Aerospace Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $3,137
Awards: G1804361226, G4795595976, G7925616758
OpenAlex ID: W4414930208