To Escape Local Minimum Problem for Multi-Agent Path Planning using Improved Artificial Potential Field-Based Regression Search Method

Artificial Potential Field (APF) method is used for the path planning through the simplest and efficient ways. Although it provides very good performances, it may fall in local minima problem, which prevents the agent to arrive its destination. The main challenge is that agents can not be able to predict local minima problem before falling it while moving for their destination in workspace. In this paper, we present a new technique that prevent multi-agent to fall in local minimum which usually happens in the case of their ways when repulsive forces are produced by obstacles and other agents. In that case, their speed becomes zero and there is no way to escape from a deadlock. In this paper, we discuss different scenarios for a workspace and show that our proposed approach gives better efficiency than other approaches.