Hierarchical path planning based on Bi‐RRT* and fractional‐order APF for vehicle obstacle avoidance in dynamic environments

Abstract

The traditional artificial potential field (APF) method in intelligent vehicle path planning has some issues. It tends to fall into local optima for the target point, and suffers from insufficient trajectory smoothness. To address these issues, a hierarchical planning method is proposed that integrates the bidirectional rapidly‐exploring random tree star (Bi‐RRT), fractional‐order APF, fuzzy logic, and a vehicle kinematic model. First, Bi‐RRT* is used to generate a globally optimal path and extract key waypoints, thereby providing a guidance for local planning. On this basis, a fractional‐order gradient is adopted to replace the traditional integer‐order potential field for local planner. The potential‐field oscillations are suppressed and the risk of local minima is reduced. Simultaneously, a fuzzy logic controller is employed to dynamically adjust the attractive and repulsive coefficients, achieving adaptive parameter matching. Finally, vehicle kinematic constraints are considered to ensure that the generated trajectory satisfies the motion feasibility of a real vehicle. Experimental results show that the proposed method effectively integrates global search and local obstacle‐avoidance capabilities. The planning success rate and trajectory executability are improved in complex dynamic environments with smooth path, short path length, and collision‐free vehicle motion.