静态障碍物下的遍历多任务目标机器人路径规划  

Traversal multi-task target robot path planning under static obstacles

作　　者：杨帆 薛亚冲[1,2] 李靖 YANG Fan;XUE Ya-chong;LI Jing(School of Electronics and Information Engineering,Hebei University of Technology,Tianjin 300400,China;Tianjin Key Laboratory of Electronic Materials and Devices,Hebei University of Technology,Tianjin 300400,China)

机构地区：[1]河北工业大学电子信息工程学院,天津300400 [2]河北工业大学天津市电子材料与器件重点实验室,天津300400

出　　处：《天津工业大学学报》2018年第4期65-71,共7页Journal of Tiangong University

基　　金：河北省自然科学基金项目(E2016202341);河北省高等学校科学技术研究项目(BJ2014013)

摘　　要：针对粒子群-遗传算法存在计算成本过高并且单一算法不能解决有障碍物存在的地图上遍历多任务目标点的移动机器人避障行走问题,提出一种分级粒子群、遗传和A*算法相结合的遍历多任务路径规划新方法.规划时,首先使用分级粒子群-遗传算法计算出执行任务的最优顺序,然后使用A*算法按照目标执行顺序进行无碰撞路径规划.该方法将遗传算法中的交叉、变异应用到粒子群算法中,提高粒子群算法的全局寻优能力和稳定性,并对粒子群进行了等级划分,不同等级的粒子在下次迭代中采用不同的操作.仿真实验证明:该算法能够规划出更优的任务目标执行顺序,并且同等目标情况下,相比于粒子群-遗传算法,迭代次数降低约25%,规划时间降低约10%.In view of the problem that the computional cost of PSO-GA algorithm is too high and the single algorithm can not guide the robot traversing the multitask target to walk on the obstacle map,a new traversing multitask targets path planning method combining the classification of PSO,genetic algorithm and A*algorithm is proposed.In path planning,the classification of PSO and genetic algorithm are used to calculate the optimal sequence of task execution in the first,and second the A*algorithm is used to avoid obstacle walking according to the target execution order.This method applies the crossover and mutation operation in the genetic algorithm to the PSO algorithm to improve the global search ability and stability,which effectively suppresses the PSO to be easily localized.Simulation results show that the algorithm can plan the better execution order of the task target,and under the same target situation,compared with PSO-GA,the number of iterations is reduced by about 25%,and the planning time is reduced by about 10%.

关 键 词：静态障碍物 分级粒子群-遗传算法 交叉和变异 A*算法 多任务目标 机器人 路径规划 

分 类 号：TP241.3[自动化与计算机技术—检测技术与自动化装置] TP301.6[自动化与计算机技术—控制科学与工程]