Evolving Neural Controllers for Collective Robotic Inspection

摘要: In this paper, an automatic synthesis methodology based on evolutionary computation is applied to evolve neural controllers for a homogeneous team of miniature autonomous mobile robots. Both feed-forward and recurrent networks can be evolved with fixed or variable network topologies. The efficacy the demonstrated in framework realistic case study collective robotic inspection regular structures, where robots are only equipped limited local on-board sensing actuating capabilities. controller solutions generated during evolutions evaluated sensorbased embodied simulation environment noise. It shown that algorithms able successfully synthesize variety novel could achieve performances comparable carefully hand-tuned, rule-based terms both average performance robustness