Introducing particle swarm optimization into a genetic algorithm to evolve robot controllers

摘要

This paper presents Swarm-Assisted Behavior Graph Evolution (SABRE), a genetic algorithm which combines elements from genetic programming and neuroevolution to develop Behavior Graphs (BGs). SABRE evolves graph structure and parameters in parallel, with particle swarm optimization (PSO) being used for the latter. The algorithm's performance was evaluated on a set of black-box function approximation problems, one of which represents part of a robot controller. We found that SABRE performed significantly better in approximating the mathematically complex test functions than the reference algorithms genetic programming (GG) and NEAT.