Characteristics of function emergence in evolutionary robotic systems dependency on environment and task

Abstract

Effective robot functions emerge from not only according to the level of system control, i.e., intelligence, but also according to the conformity between morphology, intelligence, task and environment. This paper investigates the relationship between robotic morphology and intelligence against task and/or environmental conditions and the characteristics of functions emerging in evolutionary robotic systems. The robotic system used in the study has a reconfigurable morphology and intelligence, which were designed by genetic programming. The robotic morphology is represented as a graph structure constructed from cells with motor and visual sensors, and the intelligence is a computer program defined by a parse tree. In this study, a simulation of a task in which the robot had to maintain a certain distance from an object was executed. The results clarify the importance of conformity between morphology and intelligence, and show that effective functions emerge as the result of such conformity between morphology, intelligence, task and environment.