The role of user preference in the customized control of robotic exoskeletons

Abstract

User preference is a promising objective for the control of robotic exoskeletons because it may capture the multifactorial nature of exoskeleton use. However, to use it, we must first understand its characteristics in the context of exoskeleton control. Here, we systematically measured the control preferences of individuals wearing bilateral ankle exoskeletons during walking. We investigated users’ repeatability identifying their preferences and how preference changes with walking speed, device exposure, and between individuals with different technical backgrounds. Twelve naive and 12 knowledgeable nondisabled participants identified their preferred assistance in repeated trials by simultaneously self-tuning the magnitude and timing of peak torque. They were blinded to the control parameters and relied solely on their perception of the assistance to guide their tuning. We found that participants’ preferences ranged from 7.9 to 19.4 newton-meters and 54.1 to 59.2 percent of the gait cycle. Across trials, participants repeatably identified their preferences with a mean standard deviation of 1.7 newton-meters and 1.5 percent of the gait cycle. Within a trial, participants converged on their preference in 105 seconds. As the experiment progressed, naive users preferred higher torque magnitude. At faster walking speeds, these individuals were more precise at identifying the magnitude of their preferred assistance. Knowledgeable users preferred higher torque than naive users. These results highlight that although preference is a dynamic quantity, individuals can reliably identify their preferences. This work motivates strategies for the control of lower limb exoskeletons in which individuals customize assistance according to their unique preferences and provides meaningful insight into how users interact with exoskeletons.

Human-Prosthesis Cooperation: Combining Adaptive Prosthesis Control with Visual Feedback Guided Gait

“…One potential approach is to use inverse reinforcement learning (IRL) considering the cooperative and interactive nature of the human-robot system to construct a control objective to account for goals for both humans and robot (28). Another approach is to let users adjust prosthesis control so that the prosthesis action is aligned with the intrinsic goal of human users (29).…”

Human-prosthesis cooperation: combining adaptive prosthesis control with visual feedback guided gait

“…One potential approach is to use inverse reinforcement learning (IRL) considering the cooperative and interactive nature of the human–robot system to construct a control objective to account for goals for both humans and robot [ 29 ]. Another approach is to let users adjust prosthesis control so that the prosthesis action is aligned with the intrinsic goal of human users [ 30 ]. Continued research efforts and innovative ideas are needed to determine appropriate collective human-prosthesis goals to achieve an optimized gait of individuals with amputation when walking with intelligent robotic prostheses.…”

Opportunities and challenges in the development of exoskeletons for locomotor assistance

“…These are especially important factors because it is unclear how sensitive individual wearers are to changes in metabolic cost 242 . Additional work may extend beyond metabolic cost to incorporate user preferences to rapidly adapt assistance to particular individuals and walking conditions 243 .…”