距離計測支援によるLiDAR慣性マルチビークルマッピング:退化環境への対応
Range-Aided LiDAR-Inertial Multi-Vehicle Mapping in Degenerate Environment
LiDARとIMUに加え車両間の距離計測を融合し、特徴点の法線分布から退化方向を検出して姿勢推定を補正するマルチビークルSLAMシステムを提案した。
著者: Zhe Jin, Chaoyang Jiang
分類: cs.RO
原文アブストラクト
This paper presents a range-aided LiDAR-inertial multi-vehicle mapping system (RaLI-Multi). Firstly, we design a multi-metric weights LiDAR-inertial odometry by fusing observations from an inertial measurement unit (IMU) and a light detection and ranging sensor (LiDAR). The degenerate level and direction are evaluated by analyzing the distribution of normal vectors of feature point clouds and are used to activate the degeneration correction module in which range measurements correct the pose estimation from the degeneration direction. We then design a multi-vehicle mapping system in which a centralized vehicle receives local maps of each vehicle and range measurements between vehicles to optimize a global pose graph. The global map is broadcast to other vehicles for localization and mapping updates, and the centralized vehicle is dynamically fungible. Finally, we provide three experiments to verify the effectiveness of the proposed RaLI-Multi. The results show its superiority in degeneration environments
