Liyong Wang / Beijing Information Science & Technology University
越 宋 / 北京信息科技大学
清华 苏 / 北京信息科技大学
傲 崔 / 北京信息科技大学
Ximing Zhang / China north vehicle research institute
The improved Bidirectional RRT* algorithm proposed in this paper first adopts bidirectional adaptive bias probability sampling and dynamic adjustment of step size. After generating the initial path, it uses the farthest point optimization method to 'prune' the initial path by removing redundant nodes. Finally, global smoothing using B-spline curves and local optimization using Bezier curves enhance the smoothness of the path. MATLAB simulation results indicate that compared to the RRT* algorithm, the path length is reduced by 12.27%, and the convergence speed increases by 36.39%. Compared to the Bi-RRT* algorithm, the path length is reduced by 24.46%, and the convergence speed increases by 32.09%. The improved algorithm demonstrates significant advantages in path planning efficiency, cost, and smoothness, providing reliable assurance for autonomous vehicles to quickly obtain a collision-free and smooth global optimal path.