Neural Control of an Underactuated Biped Robot

According to the fact that humans and animals show marvelous capacities in walking on irregular terrain, there is a strong need for adaptive algorithms in walking of biped robots to behave like them. Since the stance leg can easily rise from the ground, the problem of controlling the biped robots is difficult. In other words, the biped walkers have fewer actuators than the degrees of freedom. So they are underactuated mechanical systems. In this paper according to the humans and animals locomotion algorithms, the stability of an underactuated biped walker having point feet is investigated by central pattern generators. For tuning the parameters of the CPG, an effective energy based consideration is used. This robot model in single support phase is a mechanical system with one degree of underactuation. Reflexes from musculoskeletal system to CPG are modeled by tonic stretch reflex terms and nonlinear terms, where these reflexes act as a feedback loop. Also according to the dynamics of the muscles, proportional (P) controllers are used at joints for tracking