Abstract
In this study, we built an infant size robot with a vision system and a musculoskeletal system in order to investigate the effect of the interaction between a sensory system, a musculoskeletal system, and an environment on development of crawling motion. We found that the robot could learn a crawling motion by optimizing parameters of neural oscillators, using a reward function based on subjective visual information. We also found that a distance from the robot to a target object makes a difference in the convergence time of the learning process due to utilizing the subjective visual information.
