Time-Variant Gait Planning for Under-Actuated Biped Robot via Optimization

The objective of this paper is to obtain time-variant optimal gaits for under-actuated biped robot via optimization. A time-variant gait planning method based on genetic algorithm is proposed in this paper. The reference trajectories of the actuated joints are defined as polynomial functions of time t. The trajectory of the under-actuated joint can be deduced from the rotational dynamics of the biped. To get the coefficients of the polynomial functions, optimization method based on genetic algorithm is adopted and minimal torque cost is chosen as the optimize criteria. Various low torque cost gaits can be obtained by considering different constraints during optimization. Simulation results illustrate the efficiency of our method.