Abstract
Robotic systems equipped with highly responsive sensors are being investigated for use in clinical stroke assessment to overcome several important and known shortcomings of traditional tests: poor ceiling effects, poor resolution, and subjectivity. Several challenges remain for robotic metrics to be used in clinical practice. First, the robot must be safe for all users and during all assessments. Second, the robot should, through design and control, try to maintain a person's natural movement during the assessment. Finally, each metric must establish credibility via validity, reliability, and sensitivity testing. My research focuses on developing an upper-limb exoskeleton safe for people who have suffered a stroke or other neurological injury; minimizing its impedance to the user by reducing weight, friction, etc., through the use of feedback controllers; and investigating novel robotic metrics on their reliability, validity, and sensitivity.
My contributions in developing a novel adaptive admittance controller for the Bi-Lateral Upper-limb Exoskeleton for Simultaneous Assessment of Neuromuscular and Biomechanical Output (BLUE SABINO) provide an improved rendering of the desired dynamics by adapting friction and weight compared to a non-adaptive admittance controller. The improved rendering reduces the effort needed to move the robot, especially at the near-zero velocity range which occurs most frequently during a stroke assessment. The admittance controller is designed for improved safety by bounding the wrist and elbow within preset volumes. If either end-effector crosses a boundary, the robot then tries to correct it and pushes the arm back within the given region. My novel work in controller development provides an almost full range of motion and tries to maintain a person's natural movement to begin further research studies on robotic metrics for stroke assessment.