Autonomous Intermuscular Coordination and Leg TrajectoryGeneration of Neurophysiology-based Quasi-quadruped Robot

摘要

In this study, we introduce a simple self-controlled quasi-quadruped robot inspired by sensory feedback mechanisms in the hindlimb of decerebrate walking quadrupeds. The main contribution of this paper is to show that a model of the quadruped hindlimb with muscle receptors can coordinate its muscles and generate a leg trajectory without any central pattern generators. The key idea to produce coordinated patterns of limbs and muscles is based on the brainless control approach. A point of the approach is that the actuator module is embedded without any microprocessors nor a controller. Each actuator module adjusts their phases using intrinsic dynamics of them according to the reaction force received from the body-environment dynamics and, as a result, coordinated motor patterns emerge. The robot in this article is driven by artificial pneumatic muscles and controlled by only simple reflex circuits composed of mechanical air valves that switch the airflow to the muscles. Therefore, the robot needs no microprocessors or gait generators, and the only required input for the robot is a constant air-pressure which is supplied from an external air compressor. A walking experiment demonstrates autonomous gait generation, and we compare the time series of joint angles between the robot and an intact cat.