Evolutionary Behavior-Based Control of Mobile Robots

Abstract

In this chapter, we apply an ES using the statistical information of subgroups (or subpopulations) to solve a behavior-based control problem of mobile robots [68], in which a population is divided automatically into several subgroups according to a similarity of each individual, the SBMAC is used in the crossover operation, and the standard deviation calculated for each objective variable within each subgroup at a generation is applied for the mutation operator. If the above concept is introduced, then ability in searching solution is expected to be improved, and we can also avoid a tedious system design parameter setting. In the proposed method, the number of subgroups is the same as that of searching directions. The crossover decides the searching direction and the mutation decides the searching domain.