Stability control of a 6 DOF biped in the dual-support phase using fuzzy control

This work investigates the feasibility of using a fuzzy controller to stabilize a biped robot in the dual-support phase. Taking as inputs the errors of the angular position and angular velocity for each joint, the controller uses a rule-base implemented with five membership functions to compute the joint torques of a 6 DOF biped robot. The input gains were computed at run-time using the Hip and Ankle Strategy (HAS), which sets a gain value for each joint based on the relative position of the center of gravity with respect to the support area. Although the Hip and Ankle Strategy was found to increase the computational load without enhancing the actuation response, the results have clearly showed the success of the fuzzy control scheme in making the non-linear system stable, as well as in placing the robot into the most stable posture without having to lift the feet of the ground.