An identified model for human wrist movements.

We have performed tests to find the mechanical properties of the hand and muscles driving wrist flexion and extension, and have identified parameters of a model. The hand acts as a nearly pure inertial load over most of its range of motion. It can be approximated as a rigid body rotating about a single axis. Viscosity of the wrist joint is negligible. Passive elastic torques are also small, except at extreme wrist angles. We measured torque as a function of wrist angle for maximum voluntary contractions, and angular velocity as a function of load. The torque/velocity curves for shortening muscles are well approximated by a Hill equation. To measure the "series elasticity" of the muscle equivalents, we imposed step changes in torque. The series stiffness is a monotonically increasing function of the preload, or "active state", in the Hill sense. We discuss the relationship of the measured parameters to properties of isolated muscles. To see the implications of the model structure for the "inverse problem" of identifying motor control signals, we simulated four models of different complexities, and found best fits to movement data, assuming simple pulse-shaped inputs. Inferred inputs depend strongly on model complexity. Finally, we compared the best fit control signals to recorded electromyograms.