Date of Award

Fall 2016

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Biomedical Engineering

First Advisor

Beardsley, Scott

Second Advisor

Johnson, Michelle

Third Advisor

Harris, Gerald

Abstract

Cerebral palsy (CP) is a neurological condition caused by damage to motor control centers of the brain. This leads to physical and cognitive deficiencies that can reduce an individual’s quality of life. Specifically, motor deficiencies of the upper extremity can make it difficult for an individual to complete everyday tasks, including eating, drinking, getting dressed, or combing their hair. Physical therapy, involving repetitive tasks, has been shown to be effective in training normal motion of the limb by invoking the neuroplasticity of the brain and its ability to adapt in order to facilitate motor learning. Creating a device for use with Activities of Daily Living (ADLs) provides an additional tool for task-based therapy with the goal of improving functional outcome. A custom wrist orthotic has been designed and developed that assists flexion/extension of the wrist and rotation of the forearm, while leaving the hand open for the grasp and manipulation of objects. Actuated joints are driven with geared brushless DC motors on a lightweight, exoskeleton frame coupled to a passive arm that tracks positional changes within the task space. Control of actuation is accomplished with a custom mapping strategy, created from nominal movement profiles for 5 ADLs collected from healthy subjects. A simple relationship was created between position within the workspace and orientation necessary for task completion to determine needed assistance. Validation of the design subjected the device to three different conditions, including robot guidance of the limb, co-contraction of the forearm, and the use of alternate approaches to complete the task. Co-contraction and alternate approach conditions were used to simulate characteristics of impaired subjects, including rigidity spasticity, and lack of muscle control. Robot guidance achieved an average orientation error of 5° or less in at least 75% of iterations across all tasks, while co-contraction and alternate approach was able to do this in flexion/extension, but saw much higher errors in forearm rotation. Causes for performance deficiencies were attributed to lack of torque bandwidth at the motor and response delay due to signal filtering, aspects that will be corrected in the next iteration of the design.

COinS