Partial Model Based Walking Control of Quadruped Locomotion Robot with Self Renovation Control Function.

It is considered that locomotion robots are aggressive under the circumstances where human hardly work, for example, in the nuclear power plant, in the bottom of the sea and on a planet. The injury and the fault of the robot might occur frequently under those circumstances. It is very important problem that the robot can realize the walking with the fault. This is very difficult problem for biped and quadruped robot to realize a stable walking in the case that actuator or sensor is broken. And, in walking of mammal, gait pattern is generated by neural oscillator existing in the spinal cord. In the case that a lower neural system is injured, mammal realize a walking by a higher neural system. Thus, mammal has a self renovation function. In this study, in order to realize the stable walking of the quadruped robot with fault, we discuss the control method with self renovation function for the fault of the decentralized controller and the angular sensor. First, we design the centralized controller of one leg by sliding mode control for the fault of decentralized controller. Second, Sky Hook Suspension Control is applied for the fault of the angular sensor. The proposed methods are verified by 3D simulations by CAD and experiments.