Path Planning for Mobile Robots Based on the Improved DAPF-QRRT* Strategy

Abstract

The rapidly exploring random tree star (RRT*) algorithm is widely used to solve path planning problems. However, the RRT* algorithm and its variants fall short of achieving a balanced consideration of path quality and safety. To address this issue, an improved discretized artificial potential field-QRRT* (IDAPF-QRRT*) path planning strategy is introduced. Initially, the APF method is integrated into the Quick-RRT* (Q-RRT*) algorithm, utilizing the attraction of goal points and the repulsion of obstacles to optimize the tree expansion process, swiftly achieving superior initial solutions. Subsequently, a triangle inequality-based path reconnection mechanism is introduced to create and reconnect path points, optimize the path length, and accelerate the generation of sub-optimal paths. Finally, by refining the traditional APF method, a repulsive orthogonal vector field is obtained, achieving the orthogonalization between repulsive and attractive vector fields. This places key path points within the optimized vector field and adjusts their positions, thereby enhancing path safety. Compared to the Q-RRT* algorithm, the DPF-QRRT* algorithm achieves a 37.66% reduction in the time taken to achieve 1.05 times the optimal solution, and the IDAPF-QRRT* strategy nearly doubles generated path safety.