遗传算法在AGV全局路径优化中应用.pdfVIP

2008 年10 月 四川大学学报( 自然科学版) Oct. 2008 第45 卷第5 期 Journal of Sichuan University (Natural Science Edition) Vol.45 No. 5 : 2008) A V 夏 谦, 雷 勇, 叶小勇 ( , 610065) : 采用改进的链路可视图进行建模, 使用遗传算法对A V 静态已知环境进行路径规 划,并结合 A V 中的具体应用对其进行了改进. 建模过程中, 对障碍物做了边界扩展, 将 障碍物顶点作为行走节点,达到A V 行走过程中避撞的目的. 路径搜索过程中, 借鉴蚁群 算法的基于可行表的路径搜索方式到初始种群的产生过程,避免了A V 实际问题中无效路 径的产生, 提高了搜索效率. 初始种群的搜索过程中, 对死路情况做了处理, 避免了搜索停 滞.交叉过程中,采用重复点交叉的几种处理方式, 既避免了A V 搜索到断路的情况, 也较大 提高了产生最优路径的效率. 交叉变异过程中人为产生环路的做法,有利于优良 个体的产生. 将蚁群算法的正反馈优化策略引入遗传算法的多次迭代中的改进中, 加快了算法向最优解收 敛的速度. MAT LAB 仿真结果表明, 改进后的遗传算法能够快速准确地搜索到A V 的全局 最优路径. : A V; 避撞; 死路;环路; 全局路径优化;重合点交叉;遗传算法; 蚁群算法;正反馈优化 策略 : TM : A The application of genetic algorithm in global optimal path searching of AGV XIA Qian, LEI Yong, YE X iaoYong ( School of Electrical Engineering and Information, Sichuan University, Chengdu 610065, China) Abstract: The path planning is avery important part of A V control. The improved visibility graph is used to establish the environment model. The boundary of obstacles are extended in the model with the point of the obstacles as its walking nodes so that the A V can walk in a free space. The path searching schemebased on the walkable table and the positivefeedback strategy of theAnt Colony Algorithm is merged into the improve ment of the eneticAlgorithms. In the process of searching the initial generation, the problem of the blind lane is solved to avoid the searching stop. Some ways are adopted in the intercrossing process either to search the uninterrupted path or to promote the efficiency of searching out theoptimal path. The stimulation results indicate that the improved enetic Algorithm can search the global optimal path ex