Fuzzy control of bipedal running with variable speed and apex height

In this paper we propose a fuzzy-based control scheme to generate stable planar biped running gaits with variable apex height and velocity. The considered biped robot model includes five links with locked torso angles, point feet, and four actuators at the hip and knees. The controller includes two separate levels: upper-level and lower-level. The lower-level part is composed of a state machine, where the trajectory of running sub-phases and their switching time are controlled. The upper-level part includes an event-based fuzzy logic controller that is called at the apex of each flight phase. We use an offline fuzzy training process for designing fuzzy rules before controlling the robot. Fuzzy training is an iterative computational process that is repeated until convergence. Outputs of the fuzzy controller are fed into the state machine to control the running gaits. Simulation results show that the proposed control strategy generates stable gaits with controllable apex height and velocity in each step. Finally, the effects of apex height and velocity in running efficiency are investigated and optimum height is calculated as a function of running velocity.