Date of Award

Fall 2021

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Biomedical Engineering

First Advisor

Scheidt, Robert

Second Advisor

Mrotek, Leigh Ann

Third Advisor

Beardsley, Scott

Abstract

We examined a key attribute of sensory motor skill: the capability to adjust ongoing actions to correct performance errors that arise during the execution of a task. Fifteen subjects grasped the handle of a horizontal 2D planar robot that enacted a small viscous resistance to the arm movement. The subjects initiated interception movements using a prompted Go cue to quickly catch a pseudo-randomly moving target. On some trials, the robot viscosity field or the target’s motion (speed) was altered without any warning at the time of the Go cue. The viscosity could be increased which made it more difficult to move the handle or decreased which made it easier to move. Target speed could also be increased or decreased. We analyzed arm movement kinematics fitted to a sum-of-Gaussians model to determine if an error correction occurred within the interception attempt, and to quantify its timing and magnitude. We found that corrections during increasing viscosity perturbations occurred sooner (154ms) than on control trials (215ms), while perturbations during decreasing viscosity perturbations had longer latencies (272ms). By contrast, we found that visually perturbed trials had similar error correction latencies to control trials. We also found that the magnitude of the initial error correction adapted to the environmental conditions in each trial, with speed/viscosity increases eliciting more vigorous responses and speed/viscosity decreases eliciting less vigorous responses than control conditions. Finally, we found that corrections, whether they were generated internally or due to haptic or visual perturbations, were performed early in the reaching movement before sensory feedback could indicate that the target had been captured or missed. These results are consistent with models of motor control where error corrections are done in response to predicted performance rather than actual performance.

Included in

Engineering Commons

COinS