VOS-RRT: Trajectory planning for mobile robot navigation in complex pedestrian environments

Abstract

This paper introduces a trajectory planning algorithm that integrates Velocity Obstacle (VO) and Rapidly-exploring Random Tree Star (RRT*), aimed at enhancing the safety of navigation and comfort to pedestrians. Given that the original VO operates on the velocity plane while RRT* functions in the state space, a method to effectively combine these two algorithms is proposed by extending VO into the X-Y-T spatiotemporal domain and utilizing Spatiotemporal Rapidly-exploring Random Tree Star (StRRT*), which functions in X-Y-T space. The efficacy of the VO module is analyzed in three specific scenarios, and the successful integration of these algorithms to improve pedestrian comfort in complex environments is demonstrated using the Pedsim simulator.