自适应遗传算法在移动机器人路径规划中的应用

Application of adaptive genetic algorithm in robot path planning

针对传统遗传算法在路径规划时易产生不可行路径和陷入局部最优,转折次数太多等缺点,提出一种改进的自适应遗传算法。采用先验知识优化初始种群,得到不包含与障碍物相交的初始种群;设计了交叉和变异概率公式,避免了算法陷入局部最优,以提高收敛速度;在适应度函数中引入路径平滑度和路径最短作为评判标准,使规划的路径更加高效。仿真结果表明:相较于基本算法,改进算法在障碍物个数为20时,路径减少了4.2%;在障碍物为115时,路径减少了25.1%。随着障碍物不断增加,路径减少百分比呈现上升趋势,且算法迭代次数和路径中转折点个数均优于基本遗传算法。

Aiming at the shortcomings of traditional genetic algorithm in path planning, such as infeasible paths, falling into local optimum, and too many turns, an improved adaptive genetic algorithm was proposed. The initial population was optimized by prior knowledge to get the initial population which did not include the intersection with obstacles;at the same time, crossover probability and mutation probability were designed to avoid the algorithm′s falling into the local optimal solution and to improve the convergence speed. Finally the path smoothness and the shortest path were introduced at the evaluation criteria in the fitness function to make the planned path more efficient. The simulation results show that compared with the basic algorithm, when the number of obstacles is 20, the path is reduced by 4.2%;when the number is 115, the path is reduced by 25.1%. With the increasing of obstacles, the path reduction percentage presents an upward trend, and the number of iterations and that of turning points in the path of the algorithm are better than those in the basic genetic algorithm.