Date of Award
Master of Science (MS)
Beardsley, Scott A.
Scheidt, Robert A.
Schmit, Brian D.
The sensorimotor control system is a complicated system in which the neural controller uses the feedback information from sensory modalities (visual, proprioceptive, vestibular, auditory, etc.) to actuate the musculo-skeletal system in order to execute intended movements. It has been an ongoing research to decode this sensorimotor integration. The current study utilized a systems identification approach in conjunction with a one-degree-of-freedom robotic manipulandum to quantify (delays, noises, wrist dynamics and controller parameters) a simplified (linear time-invariant) model of sensorimotor control for visually guided wrist stabilization movements.
Four sensorimotor tasks were used to characterize the parameters of the sensorimotor control model. Open loop visual and proprioceptive delays along with effective feedforward delay (associated with motor processing and feedforward conduction) were estimated from subject's response to perturbation (Exp. 1) using cross-correlation analysis. Multiplicative feedforward (motor) noise was estimated by measuring the force variability in isometric torque contractions at 5 different torque levels (Exp. 2). Frequency response analysis (Exp.3 and 4) was used to obtain estimates of wrist dynamics (inertia, viscosity and stiffness), the feedback (visual and proprioceptive) gains, the controller gains (proportional, integral and derivative) and an additive sensory noise. The experimental paradigms were validated by simulating and testing the experimental task along with the sensorimotor control model in SIMULINK®. The ability of the experiments to characterize the model was tested over a range of parameter values to determine the robustness of the approach. Model performance was measured by characterizing the sensorimotor control system in 11 subjects. Variance Accounted For (VAF) by the model was used as a performance metric to compare model's response (obtained using the parameters measured for each subject in the model) with subject's performance (Exp. 5).
The proposed model of sensorimotor control contained 13 parameters, which were measured successively to study their interaction during wrist stabilization in 11 neurologically-intact subjects. The model parameters estimated for human subjects resulted in accurate predictions of hand position, with a high percentage of variance accounted for (VAF) across all subjects (78.3±3.3 %). Future studies will use these techniques to quantify how the sensorimotor control changes across tasks (tracking vs. stabilization), age and neuro-motor disabilities.