Adaptive Potential Fields for Multi-Robot Path Planning with Convergence Avoidance

Multi-robot path planning plays a crucial role in the autonomous guidance of robot fleets in various applications. The artificial potential fields approach is one of the widely used local path planning methods that offers real-time planning in continuous and dynamic environments. Nevertheless, its efficacy diminishes in multi-robot scenarios due to the low level of coordination, particularly when multiple robots risk converging on a single area, leading to traffic congestion. This paper introduces an enhanced artificial potential field method designed to minimize operation time while generating smoother paths in multi-robot scenarios. A new adaptive potential field is introduced that reduces traffic when robots converge in one area. Another potential function is proposed that prevents robots from trapping in the local minima. Also, the obstacle avoidance repulsive potentials are modified to prevent sudden changes in the robot's heading and velocity. The proposed method is developed based on a hybrid software architecture and implemented under the Robot Operating System. Real-time simulation results in the Gazebo environment depict the advantage of this method in multi-robot path planning.