动态环境中基于遗传算法的机器人路径规划

为解决动态环境中足球机器人的路径规划问题,采用栅格法对机器人工作空间进行划分,用序号标识栅格,并以此序号作为机器人路径规划参数编码,建立了以路径最短、避障为优化目标的遗传算法个体评价函数.采用轮盘赌选择、重合点交叉、多种变异结合等方法完成了遗传操作.针对遗传算法易陷入局部最优的不足,在标准遗传算法基础上加入了复原操作和重构操作,使改进后的遗传算法收敛于全局最优.仿真结果表明:该算法能够成功地在动态环境里规划出一条近似最优的路径,算法是有效的.

Soccer Robot Path-planning Based on Genetic Algorithms in Dynamic Environment

In the dynamic environment,soccer robot dynamic path planning is a difficult problem to solve.The paper proposed a method of path planning based on genetic arithmetic.The robot was supposed to move in a two-dimensional workspace with some obstacles in it.The grids were used to discrete the two-dimensional workspace.Sequence number of the grid was used to code the moving path of the robot.The sequence number was so defined that one grid corresponded to only one sequence number.This paper presented an adaptive genetic algorithm function,by which the soccer robot could move along the shortest path and avoid obstacles.And by the roulette wheel selection,coincident-point crossover and combined mutation,the genetic operation was completed.For the disadvantage of research convergence of the previous genetic algorithm,restoration operation and reconstruction operation were added to the standard genetic algorithm to make the algorithm converge to a global optimum.This algorithm was tested in dynamic environments.The simulation experiments showed that this algorithm was able to plan a better path rapidly and thus validated the effectiveness of the proposed approach.