A trajectory tracking of legged robot using decentralized control with robustness designs

A microprocessor-based decentralized sliding-mode tracking control was applied to a legged robot with two-degree-of-freedom (DOF). Under no external load, a linear discrete-time (LDT) dynamic model for every link was individually achieved by the recursive least-squares (LS) parameter estimation. An output disturbance caused by the interaction and modeling error deteriorated the system performance. In this situation, a minimax optimization of the weighted sensitivity between the output disturbance and switching surface was obtained to attenuate the effect of the output disturbance. Moreover, a suitable selection of the weighted function could reject the output disturbance of known mode. For further improving the system performance, a switching control was designed. Finally, the experiments for the legged robot with (or without) payload was arranged to evaluate the usefulness of the proposed method.