多样性指导进化算法及其在机器人路径规划中的应用

Diversity-Guided Evolutionary Algorithms and Its Application to Path Planning of Robots

通过分析及结合机器人路径规划的进化编程仿真实验发现 ,保存最优个体或淘汰最差个体都会引起进化算法早熟现象 ,而种群多样性无疑在进化算法中扮演着关键角色。虽然多样性已经用于分析算法中 ,但是很少用于指导搜索。多样性指导进化算法使用了众所周知的到平均点距离法使变异期与杂交期交替出现。多样性指导进化算法在机器人路径规划问题中展现出显著的结果 。

It is shown by analysis of evolutionary programming algorithm and experiments of it in the path planning of robots that saving the fittest or eliminating the worst in every generation of evolutionary programming algorithm may be the source of premature convergence, but population diversity is undoubtedly a key issue in the performance of evolutionary algorithms. Although various diversity measures have been used to analyze algorithms, so far few algorithms have used a measure to guide search. The diversity-guided evolutionary algorithms (DGEA) uses the well-known distance-average-point measure to alternate between phase of mutation and phase of cross. The DGEA shows remarkable result on the path planning of robots.