Nonlinear Tracking Control of an Antagonistic Muscle Pair Actuated System

When developing a machine intended for interaction with humans such as rehabilitation equipment or powered prostheses, it is useful to test in simulation prior to hardware implementation, avoiding safety hazards. Few models of the human for such purposes are seen in the literature. This paper moves toward the realization of such a model by developing a simple muscle-actuated system and related controller. The backstepping control methodology is implemented such that the controller accounts for all of the levels of intricacy presented by the Hill muscle model as an actuator, including the activation dynamics. The neural inputs are used as the actual controls. To be able to derive the control law, a modified strict-feedback form of the model is also developed. Stable tracking performance is then achieved in simulation.