An Optimization Approach for Underactuated Running Robot

In this paper, a systematic control method based on numerical optimization methods for a running robot which has a springs at a ankle joint mimicking biotic tendons is proposed. In the proposed method, periodic trajectories are generated in a combination of a numerical optimization method named PSO (particle swarm optimization) and a basic simplex method under several constraints, and the method is a systematic way to generate desired trajectories. For the realization of the motion, a feedback controller is then designed based on a linearized model around the trajectory. The validity of the method is evaluated by numerical simulations