Real-time multi-robot motion planning with safe dynamics

This paper introduces a motion planning system for real-time control of multiple high performance robots in dynamic and unpredictable domains. It consists of a randomized realtime path planner, a bounded acceleration motion control system, and a randomized velocity-space search for collision avoidance of multiple moving robots. The realtime planner ignores dynamics, simplifying planning, while the motion control ignores obstacles, allowing a closed form solution. This allows up to five robots to be controlled 60 times per second, but collisions can arise due to dynamics. Thus a randomized search is performed in the robot's velocity space to find a safe action which satisfies both obstacle and dynamics constraints. The system has been fully implemented, and empirical results are presented.