用遗传算法进行路径规划

Using genetic algorithm for path planning

本文在一种基于网络结构的并行路径规划算法的基础上，为解决该算法的全局最优问题，引入了遗传算法。由于所针对具体的问题的特殊性，本文所使用的遗传算法和通常的遗传算法在某些方面上有所不同。本文所用的遗传算法，采用了“远亲交配”的思想以获取新一代的成员，取得了较好的仿真结果。

ased on massively parallel connectionist network, to find the collision free paths,the algorithm was developed. Because of the local minimum value in this algorithm,the collision free path perhaps is not the shortest. To avoid local minimum value, the simulated annealing and some heuristic methods are used, however these methods can only ensure the collision free path, the global shortest path is not sure to be finded, so the genetic algorithm is introduced into the algorithm for its efficiency on optimization. Because of the specific property of the real question, the genetic algorithm used in this paper is different from the general genetic algorithm in some aspects. For example, in this algorithm, the problem is in continuous planning space, the coded string for the path is difficult to generate, so the path points are directly used as genetic units, and the idea 'distant relatives can pair' to create next generation is used to improve the efficiency of the algorithm. The simulation shows good results.