An Improved Potential Field-Based Probabilistic Roadmap Algorithm for Path Planning

Abstract

Probabilistic Roadmap (PRM) is one of the most important path planning techniques. It has been widely applied in mobile robot navigation for its simplicity. However, when there are narrow passages in the environment, the planning efficiency of PRM is greatly reduced. To solve this problem, this article proposes an improved potential field-based probabilistic roadmap algorithm. Making use of virtual potential field strategy, the targeting environment is represented with a quantifiable potential field map, so that the obstacle information is clearly expressed. Next, a partition-based sampling strategy is developed to improve the number of sampling points in dense obstacle areas while keeping sampling points uniformly distributed. Finally, some critical points are added with a new eight-directional detection method, thus the success rate of path planning is effectively improved. Simulation results are presented to demonstrate the effectiveness of the proposed algorithm.