基于改进粒子群算法的移动机器人多目标点路径规划  

Mobile robot multi-goal path planning using improved particle swarm optimization

作　　者：蒲兴成[1] 李俊杰[2] 吴慧超 张毅[3] 

机构地区：[1]重庆邮电大学数理学院,重庆400065 [2]重庆邮电大学智能系统及机器人研究所,重庆400065 [3]重庆邮电大学先进制造学院,重庆400065

出　　处：《智能系统学报》2017年第3期301-309,共9页CAAI Transactions on Intelligent Systems

基　　金：国家自然科学基金(51604056);重庆市科学技术委员会项目(cstc2015jcy Bx0066);重庆市教委项目(KJ1400432)

摘　　要：针对移动机器人遍历多个目标点的路径规划问题,提出了一种基于改进粒子群算法和蚁群算法相结合的路径规划新方法。该方法将目标点的选择转化为旅行商问题,并利用蚁群算法进行优化,定义了每两个目标点之间的路径规划目标函数,利用粒子群算法对其进行优化。针对粒子群算法存在的早熟现象,将反向学习策略引入粒子群算法,并对粒子群算法的惯性权重和学习因子进行改进。性能测试结果表明,改进的粒子群算法能有效避免粒子早熟现象,提高粒子群算法的寻优能力及稳定性。仿真实验结果验证了新方法能有效地实现机器人的多目标点无碰撞路径规划。真实环境下的实验结果证明了新方法在机器人多目标点路径规划的实际应用中也具有有效性。To solve the problem of multi-goal path planning for mobile robots that pass multiple goals,a new path planning method,based on improved particle swarm optimization( PSO) and ant colony optimization( ACO),is proposed. In this new method,the first step is to use an improved PSO,which has high convergence,to optimize the path between two goals among a sequence of goals. The second step is to use the ACO to obtain the shortest path for all target points. The performance experimental result demonstrates that the improved PSO algorithm can effectively avoid premature convergence and enhances search ability and stability. Simulation results show that the improved PSO algorithm can make a mobile robot realize collision-free multi-goal path planning effectively.Experiments in a real environment demonstrate that this algorithm has practical application for multi-goal path planning.

关 键 词：移动机器人 多目标点路径规划 蚁群算法 改进粒子群算法 反向学习策略 惯性权重 学习因子 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TP242[自动化与计算机技术—控制科学与工程]