A hybrid A* mobile robot path planning algorithm based on geometric-kinematic-dynamic three-level pruning

Abstract

The classical A* algorithm, when applied to mobile robot path planning, struggles to simultaneously satisfy kinematic and dynamic constraints while maintaining planning efficiency. Moreover, the mainstream L1+L2 two-level kinematic pruning hybrid A* algorithm suffers from state expansion redundancy, high computational overhead, and limited path feasibility. To address these issues, this paper proposes a geometric-kinematic-dynamic three-level pruning hybrid A* algorithm. The approach achieves progressive optimization through distance transform-based pre-pruning, motion primitive-constrained posture expansion, dynamic constraint filtering, and time-cost-optimized evaluation functions. Comparative simulation experiments demonstrate that the proposed algorithm attains a path smoothness of 0.383, comparable to that of the classical unpruned A* algorithm. Compared with the two-level pruning algorithm, the proposed method reduces the number of expanded states by 13.9% and shortens planning time by 60.6%. The algorithm significantly enhances planning real-time performance while ensuring dynamic feasibility and path quality, and exhibits favorable applicability and robustness in scenarios involving nonholonomically constrained robots.