Dynamic control allocation of a feedback linearized hybrid neuroprosthetic system

Abstract

Functional electrical stimulation (FES) can be used to restore motor function to individuals with motion impairments; however, the duration of FES usage is limited by the rapid onset of muscle fatigue. A motor-assist can be used to compensate for the muscle fatigue by sharing work load of FES. However, it is unknown how to optimally allocate control effort to the motor-assist and FES as the muscle fatigues. Further, computing an optimal control solution is challenging given the nonlinear dynamics of the human leg system. This paper uses model predictive control (MPC) to solve an optimal control trajectory for the feedback linearized musculoskeletal system, which simplifies the optimal control problem; therefore, may reduce the computational load for MPC. The feedback linearization controller was developed for the nonlinear musculoskeletal model with fatigue dynamics where FES and an electric motor torque are the inputs. Then MPC was used on the linearized musculoskeletal system to allocate control to FES and an electric motor for regulation. Simulations on a musculoskeletal model of knee extension are presented in the paper.