Open-loop tracking performance of a limb joint controlled by random, periodic, and abrupt electrical stimulation inputs to the antagonist muscle pair

The ability of the cat's ankle joint to track various input signals when controlled by electrically elicited motor unit recruitment, firing rate and antagonist muscle coactivation was examined. Pseudo-random, sinusoidal and staircase signals were used to control the soleus and tibialis anterior muscles isometrically and with a 250-g pendulum. Tracking was evaluated through cross correlation for pseudo-random and sinusoidal signals, and by rise time and steady-state error in step signals. Better tracking was obtained in isometric conditions than in load-moving conditions. Pseudo-random signals resulted in 250-ms delay between input and isometric torque output. For load-moving conditions, 340-ms and 400-ms delay in torque and angle were obtained. For sinusoids, delays decreased from 240 ms at 0.5 Hz, to 140 ms at 2 Hz in isometric conditions. Time delays for angle were between 300 and 400 ms, decreasing as frequency increased. Poor cross correlation was found for torque in load-moving conditions, because of pendulum nonlinear dynamics. Step size was not uniform in staircase trials, with steady-state errors between 9% and 39%, and rise times between 200 and 1000 ms. It is concluded that open-loop joint control results in poor tracking, presumably because it is devoid of feedback mechanisms.