Error Tolerance in Wearer-Robot Systems

Project Summary
Human's motor control system adopts various control mechanisms to tolerate internal or external error/noise, so that
humans can maintain consistent task performance in an extremely robust way. Whether or not the same control principles
can be employed by wearable robots (such as robotic prostheses and exoskeletons) to enhance the robustness of wearer-
robot systems remains an open but paramountly important question. Answers to this question can lead us to understand the
processes underlying wearer-robot interactions and make advanced wearable robots robust, safe-to-use, and acceptable by
the wearers. Our long-term goal is to achieve seamless wearer-robot integration for movement augmentation and clinical
translation of knowledge and technology in wearer-robot systems to improve the quality of life of individuals with motor
deficits. Specifically, the objective of this proposal is to investigate novel error-tolerant mechanisms, inspired by human
motor control principles, to improve the robustness and safety of powered transfemoral prostheses. By (1) systematically
exploring the stability response of amputee-prosthesis system to imposed prosthesis errors and (2) mimicking how humans
explore the control space and tolerate internal or external errors for task performance, we will demonstrate a new bio-
inspired error-tolerant concept for robust control of robotic prostheses.
Guided by strong preliminary design and study, our research objective will be accomplished by pursuing three
specific aims: Aim 1) systematically determine the effects of imposed prosthesis errors on objectively measured walking
stability of amputee-prosthesis systems, Aim 2) systematically determine the effects of imposed prosthesis errors on
perceived walking stability in amputees, and Aim 3) demonstrate the capacity of robust lower limb prosthesis control by
mimicking human motor control mechanisms.
The goal of Aim 1 and Aim 2 is to comprehensively understand the consequences of prosthesis errors and the responses of
amputee-prosthesis systems to these errors, which is an existing knowledge gap that hinders the development of robust
robotic prosthesis controller. By quantifying and correlating the stability measures (both perceived and biomechanically
defined indices), we will map a manifold surface that can truly reflect the responses of amputee-prosthesis system to
prosthesis errors. In Aim 3, we propose to translate the knowledge learned in Aim 1 and 2 together with the human motor
control theories into error-tolerant mechanisms for powered prostheses. Guided by Minimal Intervention Principle and the
response manifold obtained in Aim 1 and 2, error-tolerant mechanisms will be designed to accurately detect prosthesis errors
that lead to perceived instability and effectively correct errors via a forward model, and therefore in turn enhance the stability
in amputee-prosthesis systems. The success of this proposal can provide theoretical foundations for the development of
technologies that can improve amputees' safety in using robotic prostheses, enhance acceptance of these advanced devices,
and improve the quality of life of amputees.