Voluntary movements for robotic control

A neuromuscular-like nonlinear mathematical model for controlling robotic limbs is described. The model consists of two muscle-reflex models representing a pair of muscles acting as an agonist and an antagonist in moving a load. The responses of the model have been compared with experimental data from human wrist movements in order to identify command signals in the form of a series of rectangular pulses for effecting the measured movements with two different loads. These results can be used to obtain several empirical rules for muscle modulation control. The proposed control strategy can also compensate for unexpected disturbances, which is an essential capability for compliance control.<>