An Improved Particle Swarm Optimization Method for Motion Planning of Multiple Robots

Multi robot motion planning is a challenging problem in the robotics field due to its complexity and high computational costs induced by the number of robots. In this paper, a new heuristic method is presented for solving this problem through a decentralized approach with global coordination. The method is based on a new improved variant of the Particle Swarm Optimization (PSO) metaheuristic, which serves as a global planner. Alternatively, for local planning and avoiding obstacles in narrow passages, the Probabilistic Roadmap Method (PRM) is employed. The global and local planners act sequentially until all robots reach their goals. The algorithm iteratively and simultaneously minimizes two main objectives, shortness and smoothness of the paths. The proposed algorithm is simulated and compared with the standard (basic) PSO, as well as the standard Probabilistic Roadmap methods. The experimental results show a meaningful advantage of the new method regarding computation-al time and path quality.