Model predictive control of human elbow joint movement

Abstract

Human movement can be divided into two phases: planning and execution. During the first phase, the human plans the movement trajectory by satisfying minimize jerk (or maximum smoothness) criteria. During the execution phase, the neural system uses the equilibrium point trajectory as control signal (muscle activation signal) to trace the planned trajectory. This paper builds a model of the neural musculoskeletal system by integrating the springlike feature of muscle. Based on theories of equilibrium point and model predictive control, the equilibrium point trajectory is derived. The simulation results show that the model predictive controller can transfer the centrally planned motor intention into activation signals of muscles by producing control signal (equilibrium point trajectory). Excited muscles produce torque around the elbow joint and drive the limb to the terminal position. The equilibrium point trajectory shows a similar shape to the planned trajectory and the elbow joint velocity trajectory shows a ballistic shape.