Autonomous decentralized gait control of a 6-legged walking robot using immune networks (gait generation through reinforcement learning)

Biological systems in living organisms can be regarded as the ultimate among distributed information processing systems and have been expected to provide feasible ideas to engineering fields, especially robotics. Among these systems, the immune system plays a role in protecting living organisms against countless foreign substances in order to cope with a dynamically changing environment. This ability is exactly what we need for an autonomous mobile robot. Therefore, in this paper, we focus on the immune system and apply it to a gait control problem of a 6-legged walking robot. We aim at realizing an action selection mechanism through a decentralized learning method inspired from the immune network architecture. The validity of our proposed method is confirmed by simulations.