改进RRT算法的全向机器人路径规划  

Omnidirectional Robot Path Planning Based on Improved RRT Algorithm

作　　者：牛秦玉[1] 高乐乐 闫朋朋 NIU Qin-yu;GAO Le-le;YAN Peng-peng(Xi'an University of Science and Technology,Shaanxi Xi'an 710000,China)

机构地区：[1]西安科技大学,陕西西安710000

出　　处：《计算机仿真》2024年第1期473-478,507,共7页Computer Simulation

基　　金：钻孔救援用变结构探测机器人设计理论与自主越障/避障规划策略(52174149)。

摘　　要：针对快速扩展随机树(RRT)算法在进行全局路径规划时生成路径时间长,长度长,在采样时缺乏目标导引性。所以改进了一种目标偏向采样策略,通过引入由初始点,终点,障碍物和随机点构成的虚拟力场,促使随机采样导向目标点。对于障碍物远近的影响,设定障碍物影响的阈值,引入了自适应系数λ,在扩展步长方面,加入了角度约束,根据偏向目标点角度,选择不同的扩展步长。引入冗余节点剔除策略,缩短路径长度,最后加入贪心策略,减少不必要采样节点的个数。首先,在MATLAB2018a仿真进行实验对比,其次在实际环境中进行实验。实验结果显示,在路径搜索时间,拐点数量以及路径距离等均有明显的提高。The fast expanding random tree(RRT)algorithm takes a long time and length to generate the path in the global path planning,and lacks the target guidance in the sampling.In this paper,a target biased sampling strategy is proposed,which promotes random sampling to target points by introducing a virtual force field composed of initial points,end points,obstacles and random points.For the influence of the distance and proximity of obstacles,the threshold of the influence of obstacles was set,the adaptive coefficient is introduced,the angle constraint was added in the expansion step,and different expansion steps were selected according to the angle of the target point.Aiming at the path length redundancy of the traditional RRT algorithm in the global path planning,the redundant node elimination strategy was introduced to shorten the path length.Finally,the greedy strategy was added to reduce the number of unnecessary sampling nodes.Firstly,the simulation was carried out in matlab2018a,and then the experiment was carried out in the actual environment.The experimental results show that the path search time,the number of inflection points and the path distance are significantly improved.

关 键 词：路径规划 目标偏向采样 角度约束 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]