分布式协同多机器人多任务目标遍历路径规划  

Distributed cooperative multi-robot traversing multi-task target path planning

作　　者：李靖 杨帆 LI Jing;YANG Fan(School of Electronic and Information Engineering,Hebei University of Technology,Tianjin 300401,China;Tianjin Key Laboratory of Electronic Materials and Devices,Hebei University of Technology,Tianjin 300401,China)

机构地区：[1]河北工业大学电子信息工程学院,天津300401 [2]河北工业大学天津市电子材料与器件重点实验室,天津300401

出　　处：《天津工业大学学报》2020年第6期68-75,共8页Journal of Tiangong University

基　　金：天津市自然科学基金资助项目(18JCYBJC16500);河北省自然科学基金资助项目(E2016202341)。

摘　　要：针对群智能优化算法易陷入局部最优且单一算法不易解决障碍物空间中多机器人多任务目标遍历的问题,提出一种分布式协同多机器人多任务目标遍历路径规划策略。首先,通过K-Means聚类算法对多任务目标进行分类,随后运用改进的灰狼优化算法求解每类任务目标的最优遍历顺序,其中改进的灰狼优化算法引入余弦收敛因子以平衡全局搜索与局部开发的能力,引入布谷鸟搜索算法优化种群更新位置,最后在类内根据遍历顺序运用A*算法避障路径规划。每类的任务目标遍历路径规划的集合即为整个系统的多任务目标遍历路径。仿真实验表明:在规划多任务点遍历路径时,改进的灰狼优化算法比传统灰狼优化算法求解的路径长度缩短了5.08%,且适应度曲线收敛更快、算法稳定性更高;在规划避障路径时,A*算法比模糊逻辑法与RRT法求解的路径长度分别缩短了22.4%、9.8%,同时验证了分布式协同多机器人多任务目标遍历路径规划算法的可行性。In order to solve the problem that swarm intelligence optimization algorithm is easy to fall into local optimum and the single algorithm can not guide multi-robot traversing the multi-task target to walk in obstacle space,a path planning strategy for multi-task target traversal of distributed cooperative multi-robot is proposed.Firstly,the multi-task targets are classified by means of K-Means clustering algorithm,and then the improved gray wolf optimizer algorithm is used to solve the optimal traversal order of each task target,in which,cosine convergence factor is introduced to balance the ability of global search and local development,and Cuckoo search algorithm is introduced to optimize the update position of population to improve gray wolf optimizer algorithm.Finally,A*algorithm is used to avoid obstacles in the class according to the traversal order.The set of traversal path planning for each multi-task target is the path of the whole system.The simulation results show that when traversing the multi-task points,the path length of the improved gray wolf optimizer algorithm is 5.08%,shorter than that of the traditional gray wolf optimizer algorithm,and the fitness curve convergence is faster and the algorithm is more stable.When planning the obstacle avoidance path,the path length of A*algorithm is 22.4%and 9.8%shorter than that of fuzzy logic algorithm and RRT algorithm,and the feasibility of distributed cooperative multi-robot traversing multi-task target path planning is proved.

关 键 词：分布式协同 多机器人 路径规划 多任务目标 K-MEANS聚类 布谷鸟搜索算法 灰狼优化算法 A*算法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]