Velocity obstacle-based trajectory planner for anthropomorphic arms

Abstract

The Velocity Obstacle algorithm is one of the most popular and studied decentralized trajectory planning methods for multi-agent systems moving in dynamic environments. It has been successfully used in a multitude of real and simulated scenarios for collision-free maneuvers of ground and aerial mobile robots, proving to be versatile, adaptable, and efficient even with large numbers of holonomic and nonholonomic autonomous agents. In this paper, we address the problem of adapting Velocity Obstacles for planning collision-free trajectories for anthropomorphic arms. We show MATLAB® simulations and experimental tests on a real UR5 robotic arm by Universal Robotics in the presence of moving obstacles. Finally, we compared the collision-free trajectories of the proposed planner to those obtained by using Artificial Potential Fields: our approach gives better results in terms of both smoothness and time employed for reaching the target position.