Publications
N. Thomas,
Neural Efficiency of Haptic Feedback and Autonomous Control in Upper-limb Prostheses, (2022),
[Doctoral Dissertation, Johns Hopkins University]. (Abstract,links)
Abstract
Upper-limb absence renders everyday tasks extremely difficult or even impossible to complete. Although clinical myoelectric prosthetic hands can partially restore function, they lack the dexterity inherent to human sensorimotor control. Current noninvasive approaches to addressing this issue include providing the user artificial haptic feedback or imbuing the prosthesis with autonomous control. However, both haptically guided volitional control and reflexive control are required to truly replicate human sensorimotor control. Inspired by biology, this work evaluates the utility of combining haptic feedback and autonomous control in a myoelectric prosthesis. The research carried out in this thesis presents a systematic approach to investigating the functional and neurophysiological impact of a sensorimotor-inspired prosthesis control scheme. Unique to this approach is the utilization of task performance and neural imaging cognitive load measures to assess neural efficiency, the mental effort required to achieve a certain level of performance. We first compare the neural efficiency of a prosthesis featuring haptic feedback with the standard prosthesis and the biological hand. Here, functional near infrared spectroscopy (fNIRS) is used to measure prefrontal cortex activity during a dexterous object-discrimination task. Next, we develop and assess a hybrid system comprised of haptic feedback and autonomous reflexive controllers. This system is compared to the standard prosthesis in a dexterous pick-and-place task conducted without direct vision. Finally, using fNIRS, we assess the neural efficiency of combining haptic feedback and autonomous volitional control in a haptic shared control scheme. We compare this system to the standard prosthesis and the prosthesis with only haptic feedback in a dexterous grasp-and-lift task. Results from these three investigations demonstrate that haptic feedback decreases cognitive load compared to a standard prosthesis, complements autonomous reflexes in different aspects of a dexterous task, and optimizes neural efficiency when used in a shared control scheme. Our findings highlight the benefits of haptic feedback combined with autonomous control, and the efficacy of fNIRS in assessing cognitive load during prosthesis use. This work represents the first look at a sensorimotor-inspired hybrid system with both performance and neurophysiological evaluations, and is thus well-situated to influence future improvements to bionic limbs and human-robot interaction more broadly.
Links
http://jhir.library.jhu.edu/handle/1774.2/67236