Abstract
To address the issues of low sampling efficiency, slow convergence speed, and poor environmental adaptability of the Rapidly-Exploring Random Tree Star (RRT*) algorithm in complex environments with dense obstacles and narrow passages, this paper proposes a bidirectional rapidly-exploring random tree star path planning algorithm based on Halton sampling and multi-strategy expansion (HGSM-BI-RRT*). First, high-quality sampling points are generated through probabilistic bridge sampling based on Halton sequences to optimize global sampling and improve the traversability of narrow passages. Secondly, a three-level expansion strategy guided by dynamic goal switching is designed, which integrates window-adaptive goal-biased direct expansion, sector-based layered candidate expansion, and random guaranteed expansion, balancing search speed and obstacle avoidance capability. Additionally, anchor-based greedy rewiring is employed for local path optimization to reduce path cost. Finally, cubic B-spline interpolation is implemented to achieve path smoothing, outputting a smooth and safe optimal path. Mobile robot experimental results demonstrate that HGSM-BI-RRT* effectively enhances the overall performance of the algorithm. In practical dense and narrow passage environments, compared with RRT*, BI-RRT*, and BI-APF-RRT* algorithms, the proposed algorithm reduces path length by a maximum of 33.9%, 29.4%, and 23.4%, shortens pathfinding time by a maximum of 60.3%, 52.9%, and 50.9%, and decreases the number of turns by a maximum of 72.7%, 62.5%, and 50.0%, respectively, validating the feasibility and superiority of the algorithm.
Get full access to this article
View all access options for this article.