NSF Award Search: Award # 2110214 (2024)

• Awards
• 
• Search Awards
• 
• Recent Awards
• 
• Presidential and Honorary Awards
• 
• About Awards
• How to Manage Your Award
• Grant General Conditions
• 
• Cooperative Agreement Conditions
• 
• Special Conditions
• 
• Federal Demonstration Partnership
• 
• Policy Office Website

Collaborative Research: Legible Co-Adaptation of Wearable Devices for As-Needed Assistance of Arm Motion

ABSTRACT

For people who have difficulties moving their arms or grasping objects, an exoskeleton can be used to augment a person?s capabilities and, in some cases, to provide rehabilitation. An arm exoskeleton can take the form of a brace worn from the hand to above the elbow, with motors to assist movement and to provide additional strength. Existing commercial exoskeleton systems have software with control parameters that can be adjusted to an individual?s capabilities, needs, and comfort, but these settings are not modified automatically while the device is being used. In some cases, a person could fatigue over the course of a long task, thus requiring the exoskeleton to help more at the end of a task than at the beginning. In other cases, a person might gain more ability as they are assisted in movements over days or weeks, allowing the exoskeleton to start providing less assistance. The motivating objective of this research is to enable co-adaptation of an arm-worn exoskeleton, where the control parameters of the exoskeleton adapt over time to the person?s changing capabilities, generating synergistic coordinated motion between the human-exoskeleton team. This project promotes the progress of science and advances the national health, prosperity and welfare through the development of an adaptive exoskeleton controller designed to provide as-needed motion assistance based on the user's changing level of muscular fatigue and capabilities. The research considers augmenting the abilities of healthy participants, which aligns with reducing musculoskeletal risk for an industrial or military application, as well as motion assistance for older populations, which can aid with sensorimotor deficits due to aging. The project addresses fatigue detection, legible notifications, and device assistance to enable synergistic physical and cognitive fit between the human and intelligent exoskeleton machine. This convergent research brings together theories from computer science, human factors, robotics, physical therapy, and kinesiology to advance a fundamental understanding of human-robot interactions. The project also includes public engagement through STEM activities for middle school girls, as well as through activities for life-long learning for older adults.

The function of the co-adaptive algorithm can be expressed using principles motivated by dynamical systems theory. The algorithm will adapt the exoskeleton controller at the parameter dynamics level (a time scale describing parameters associated with completing the task) for a specified graph dynamics (a longer time scale describing the connectivity architecture representing the system). The algorithm encourages stability of the graph dynamics using a legibility scheme. The project team will use the commercially available Myomo MyoPro powered orthosis as a platform to evaluate co-adaptation of a human-exoskeleton team. The project team plans a control legibility scheme that will convey a series of notifications to the user to inform her/him about what the MyoPro is doing. The legibility scheme is designed to align with enabling a calibrated trust of the user with respect to the device and to prevent bifurcations in the graph dynamics. The specific objectives of the research are: (1) to incorporate measures of muscle fatigue in a legible co-adaptive exoskeleton controller; (2) to evaluate the hypothesis that a co-adaptive controller can improve human-exoskeleton task performance in the presence of fatigue in a healthy young adult population; and (3) to examine the generalizability of the co-adaptive controller to an older adult population. The primary task across all studies is a pick-and-place task where participants grasp an object (e.g., a book, pen, basket, cup) from a table and place it on a bookshelf, then return the object to the table. The use of different object sizes and weights increases the number of grasp types in the dataset, as well as the motor control strategies required, providing increased generalizability of the algorithms across grasping tasks. Participants will perform concurrent cognitive tasks that will be used to assess cognitive load during the primary pick-and-place task. To increase the rate of fatigue, the primary task is alternated with a fatiguing task. Results from these studies will advance the usability of exoskeleton systems by accounting for natural forms of human adaptation within the exoskeleton control policy, enabling exoskeletons to be applied for longer-duration applications that support aging in place and the mitigation of injury.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

Please report errors in award information by writing to: awardsearch@nsf.gov.