Phillip Grice
Robotics Ph.D. Candidate
Dept. of Biomedical Engineering
Georgia Institute of Technology

Robotics PhD Defense Presentation
Title: Assistive Mobile Manipulation for Users with Severe Motor Impairments
Date: June 12th, 2017 (Monday)
Time: 10:00 AM EDT
Location: 3115 (McIntire Conference Room), BME Whitaker Building

Committee
Dr. Charles C. Kemp, Advisor (Dept. of Biomedical Engineering, Georgia Institute of Technology)
Dr. Karen Feigh (School of Aerospace Engineering, Georgia Institute of Technology)
Dr. Andrea Thomaz (Dept. of Electrical and Computer Engineering, University of Texas at Austin)
Dr. Lena Ting (Dept. of Biomedical Engineering, Emory University)
Dr. Randy Trumbower (Dept. of Rehabilitation Medicine & Dept. of Biomedical Engineering, Emory University)

Abstract
Mobile robots that can physically manipulate their environments have the potential to help persons with motor impairments perform a variety of tasks, including self-care tasks. Such assistive mobile manipulators (AMMs) could increase independence and reduce the burden placed on caregivers. However, general-purpose AMMs are complex systems with many sensors and actuators, and so are often difficult for non-experts to operate. Through user-centered research, we have developed software enabling the use of a general-purpose robot as an accessible AMM to address these challenges. This process has included a long-term collaboration with an individual with severe quadriplegia, Henry Evans, and his wife and primary caregiver, Jane Evans. 

We evaluated our system with 15 inexperienced users with severe motor impairments from around the country. These participants, having a variety of sources of impairment and using a variety of accessible computer interfaces, operated the robot remotely to perform both a standardized clinical manipulation assessment and a simulated self-care task. When operating the robot, they achieved a significant improvement over the use of their own limbs, and their improvement on the clinical assessment exceeded a conservative estimate of the minimal clinically important difference. Furthermore, their feedback provided support for the likelihood of technology adoption, despite slow task completion. We also evaluated our system during a seven-day deployment in Henry Evans' home. During this study, Henry operated the robot for over 22.5 hours in 16 separate sessions, successfully performing more than 18 distinct tasks, some numerous times. Our study demonstrates the broad capability of general-purpose AMMs and provides a baseline for teleoperation performance