A Study on Evolutionary Design in Balancing Morphology and Intelligence of Robotic Systems
Kohki Kikuchi and Fumio Hara
Department of Mechanical Engineering, the Science University of Tokyo, 1-3 Kagurazaka, Shinjuku-ku, Tokyo, 162-8601, Japan
This paper deals with an evolutionary design of a robotic system and structural characteristics of its morphology and intelligence. Morphology affects robotic behavior through physical interaction with the environment, and intelligence determining the behavior strongly depends on the morphology design such as visual sensor position, touch sensor position, body shape, and body size. We propose an evolutionary method that designs both morphology and intelligence of the robotic system concurrently using genetic programming. This design offers robotic systems able to execute a given task through interaction with the environment such as gravity, collision arid friction. Robotic systems possess high performance for a given task through harmonizing morphology and intelligence against changing conditions of a real world. We found that the complexity of morphology is correlated with robustness in robotic behavior, and physical dimensional inconsistency in intelligence is correlated with fitness in evolutionary design.
Copyright© 2000 by Fuji Technology Press Ltd. and Japan Society of Mechanical Engineers. All right reserved.