Simultaneous design of the sensory morphology and controller of mobile robots

摘要: This paper proposes a method for automatic design of the sensory morphology mobile robot. The proposed employs two types adaptations, ontogenetic and phylogenetic, to optimize In adaptation, reinforcement learning searches optimal policy, which is highly dependent on morphology. phylogenetic genetic algorithm used select morphologies with robot can learn tasks faster. Our was applied line-following We performed simulation experiments compare solution hand-coded results revealed that our outperformed in terms following accuracy speed, although had fewer sensors than one. also built physical using solution. experimental this its effectively © 2010 Wiley Periodicals, Inc. Electr Eng Jpn, 172(1): 48–57, 2010; Published online InterScience (www.interscience.wiley.com). DOI 10.1002/eej.20965