Multi-objective stochastic search based on network topology and its application to evolutionary robotics

对于由多个目标函数组成的多目标优化问题，获取Pareto最优集是最重要的。从这个角度来看，多点搜索方法由于能够通过一次运行提供Pareto最优候选者而引起了人们的关注。在本研究中，我们考虑了基于搜索点网络的随机多点搜索方法进行多目标优化，以定义搜索点之间的信息共享范围。此外，我们还讨论了在机器人世界杯足球仿真中足球机器人运动采集的应用。

To obtain Pareto optimal set is primary important for the multi-objective optimization that consists of plural objective functions. From this point of view, multipoint search methodology has attracted attention, since it is able to provide the Pareto optimal candidates by single run. In this study, we have considered stochastic multi-point search methodology for multi-objective optimization based on the search point network to define the information sharing range among search points. In addition, we discussed an application to motion acquisition of the soccer robot in the RoboCup soccer simulation.