Neuromuscular control model of the arm including feedback and feedforward components.

A neuromusculoskeletal model of the human arm was developed which contains both feedforward and feedback control, and thereby accounts for motor control of fast movements as well as interaction with external forces. The feedforward control component forms an approximate representation of the inverse dynamics of the arm and its interaction with the environment. The feedback control component compensates for errors in the representation of the inverse dynamics and for unexpected forces acting on the arm. Moreover, the control system provides a solution for the redundancy of the muscles. The system performance is adapted in a learning procedure according to a specified goal function. It is shown in the paper that good control of the nonlinear musculoskeletal model and neural control signals which are similar to electromyographic (EMG) data, are attained. The response of the arm to external forces is analysed and compared with experimental data on arm impedance.