【摘要】 针对双向快速搜索随机树(RRT-Connect)算法在机械臂路径规划中存在的采样随机性强、收敛速度慢、生成路径冗长等问题，提出了一种改进的RRT-Connect算法。首先，采用动态概率阈值的目标偏置策略，避免采样过程陷入局部极小值；其次，对新节点引入预定义参数深度的重选父节点环节，并对每次迭代生成的新路径和两树连接处设置转角约束，优化路径质量；最后，对初始路径进行剪枝优化，删除冗余节点，利用二阶贝塞尔曲线进行路径平滑处理，使其更符合机械臂的动力学特性。实验结果表明，改进算法的场景适应能力更强，路径质量和搜索效率更高。更多还原

【Abstract】 Aiming at the problems of bidirectional rapidly-exploring random tree(RRT-Connect) algorithmin robotic arm path planning, such as strong sampling randomness, slow convergence speed, generating verbose paths, an improved RRT-Connect path planning algorithm was proposed. Firstly, employing a target bias strategy with dynamic probability thresholds to prevent the sampling process from falling into local minima. Furthermore, introducing a predefined parameter for depth in the reselecting parent nodes process for new nodes, while setting corner constraints for the newly generated paths and connections between the two trees, enhances the quality of the path. Finally, pruning optimization is applied to the initial path by removing redundant nodes, and second-order Bézier curves are utilized to smooth the path, making it more in line with thekinematic characteristics of the robotic arm. The experimental results indicate that the improved algorithm demonstrates stronger adaptability to different scenarios, with higher pathquality and search efficiency.更多还原