基于多策略引导和自适应扩展的RRT*三维路径规划算法OA
RRT* path planning algorithm based on multi-strategy guided and adaptive exploration in three-dimensional environments
针对复杂三维环境下路径规划的安全性和高效性等问题,提出了一种基于多策略引导和自适应扩展的快速探索随机树算法(modified & multi-strategy rapidly exploring random tree*,MR-RRT*).首先,采用双向树扩展机制以减少迭代次数;其次,通过设定动态概率函数来自适应调整目标偏向概率,使得扩展树的生长能适应不同阶段的需求;同时,引入改进的人工势场法以引导树的扩展路径,考虑当前采样点和目标点距离的排斥力以解决局部最小问题,并通过调整斥力参数来调节斥力的强度.在路径后优化阶段,采用贪婪剪枝算法去除冗余节点,并利用B样条曲线对路径进行平滑处理,生成连续可行的最终路径.实验结果表明,MR-RRT*算法与几种经典的RRT*改进算法相比,平均节点数量减少了 54.25％～85.73％,平均安全距离增加了 74.75％～93.18％,而且在平均路径长度和时间消耗上也具有优势,显著提高了路径规划的效率、鲁棒性和安全性.
This paper developed an MR-RRT*algorithm to improve path safety and efficiency in complex three-dimensional environments.The algorithm adopted a bidirectional expansion mechanism to reduce the number of iterations.It used a dynamic probability function to adaptively adjust the target bias for different exploration stages.An improved artificial potential field method guided tree expansion.The repulsive force between the sampling and goal points helped avoid local minima.The re-pulsion strength was controlled through parameter adjustment.In the post-optimization phase,a greedy pruning algorithm re-moved redundant nodes.B-spline smoothing generated a continuous and feasible final trajectory.Simulation results show that,compared with several classical RRT*variants,the MR-RRT*algorithm reduced the average number of nodes by 54.25％～85.73％and increased the average safety distance by 74.75％～93.18％.It also achieves shorter path length and lower com-putation time.Overall,the algorithm improves the efficiency,robustness,and safety of path planning.
于现凯;刘德胜;张燕
佳木斯大学信息电子技术学院,黑龙江佳木斯 154007佳木斯大学信息电子技术学院,黑龙江佳木斯 154007佳木斯大学信息电子技术学院,黑龙江佳木斯 154007
信息技术与安全科学
路径规划快速探索随机树算法动态概率函数人工势场法贪婪算法
path planningRRT*dynamic probability functionartificial potential field methodgreedy algorithm
《计算机应用研究》 2026 (6)
1740-1745,6
黑龙江省优秀青年教师基础研究支持计划(YQJH2023219)佳木斯大学校级重点项目(SZDYF202302)
评论