基于结构感知采样的RRT机械臂路径规划OA

To address the issues of poor path quality,blind tree expansion,and slow convergence of the rapidly-exploring random tree(RRT)algorithm in three-dimensional path planning,an improved structure-aware sampling algorithm named hot-segment feedback directed RRT(HFD-RRT)is proposed.The algorithm adopts hot-segment feedback as its core mechanism,where key hot segments are identified based on path curvature and obstacle density,and a heatmap is dynami-cally updated to guide tree expansion.On this basis,a multi-strategy sampling framework is constructed to achieve a dynamic balance between global exploration and local optimization.Moreover,an environment-aware cost function is introduced to optimize parent node selection,and a dynamic step-size adjustment strategy is integrated to enhance the struc-tural adaptability of the search tree.Simulation results demonstrate that HFD-RRT consistently outperforms GB-RRT,RRT*,Informed-RRT*,and AE-RRT* in quantitative metrics including path length,planning time,and path smoothness.Further simulations on a UR5 robotic manipulator verify the geometric executability of the planned paths.