Approach to global path planning and optimization for mobile robots based on multi-local gravitational potential fields bias-P-RRT*

Abstract

The sampling-based method has strong environmental adaptability and probability completeness, providing an effective solution for mobile robot path planning. However, the conventional rapidly-exploring random trees (RRT) algorithm often presents slow convergence and inefficient search paths. In this sense, this paper proposes a mobile robot path planning and optimization algorithm based on P-RRT* that incorporates multi-local gravitational potential fields and bias sampling, i.e., multi-local gravitational potential fields Bias-P-RRT* (MLGPFB-P-RRT*). The algorithm adds a local gravitational field between the starting point and the target point to better guide the direction of random tree growth, and directly connects the center of the last local gravitational field to the target point to accelerate the convergence of the random tree at the target point. Meanwhile, the introduction of bias sampling based on local potential fields to optimize the generation quality of random points, thereby improving the generation position of new nodes and reducing the randomness of sampling for mobile robots in the workspace. Then, a collision detection method between sampling nodes and obstacles was developed, which can quickly determine the feasibility of the sampling path. Finally, the generated path is optimized and smoothed through pruning optimization and quadratic B-spline function. A series of simulation studies and mobile robot experiments demonstrate the superior performance of the proposed algorithm.