A Method for Map Construction in Degraded Environments of Subway Tunnels

Abstract

To meet the demand of high-precision point-cloud map construction under the scenes of degraded long subway tunnels, this paper proposes an offline map construction method based on LiDAR and inertial measurement sensors. The method consists of a tight coupling front-end odometry based on error-state Kalman filters and back-end optimization based on factor graphs. The front-end odometry predicts using the results of inertial computation and updates the filter based on the residual constraints of the current frame's points to the local map plane. The global pose graph is constructed based on inter-frame odometry and other constraint factors, followed by a smoothing optimization process to build the map. Results from multiple experiments in tunnel environments of rail transit demonstrate that the map construction maintains accurate poses without degradation. The trajectory errors of multiple data map construction are less than 0.1 meters, and the trajectory consistency meeting the precision requirements of active collision avoidance for trains. This validates the effectiveness of the proposed method.

Keywords

• map construction
• tight coupling
• graph optimization
• rail transit
• tunnel