A HBMO-based batch beacon adjustment for improving the Fast-RRT

Abstract

Fast-RRT improves on the original rapidly-exploring random trees (RRT) by incorporating two main stages: improved-RRT and fast-optimal. The improved-RRT stage enhances the search process through fast-sampling and random steering, while the fast-optimal stage optimizes the path using fusion and path arrangement. However, path fusion can only be optimal when the newly found path is unique and different from previous paths. This uniqueness rarely occurs in cases with narrow corridors, so path fusion only provides suboptimal conditions. To address this, the study explores using honey bee mating optimization (HBMO) to optimize or replace the fusion stage. HBMO helps determine new beacon coordinates, which are nodes between the start and goal points along the path, through a batch beacon adjustment approach. The results show that integrating HBMO into FastRRT improves its optimality, with a 21.85% reduction in path cost and a 5.22% decrease in completion time across environments with varying difficulty levels. This hybrid algorithm outperforms previous methods in terms of both path optimality and convergence rate, demonstrating its effectiveness in enhancing Fast-RRT’s performance.