CHS: Medium: Collaborative Research: Collaborative Online Learning and Control for Motor Prosthesis

点击翻译按钮获取中文摘要

It is estimated that approximately 5.4 million people in the United States live with some form of paralysis, defined as a central nervous system disorder resulting in difficulty or inability to move the upper and/or lower extremities. Many paralyzed individuals consider restoration of lost basic motor functions such as grasping and walking as important abilities that could improve their quality of life. The goal of this project is to develop and evaluate advanced machine learning algorithms that enable quadriplegic individuals to control a robotic hand with heterologous muscles (that is, muscles not typically involved in moving the limbs; for example, muscles of the neck). To enable this research, prospective algorithms will be initially evaluated in normally-enabled individuals. The most promising algorithms will subsequently be evaluated in quadriplegic individuals. This research is a first step toward providing benefit to the paralyzed community by creating pathways toward the development and commercialization of functional motor prosthetic systems. Paralyzed individuals can be trained to use the system by planning the movements in their minds, much like moving their natural limbs. Success of this research could lead to a significant advance in improving function and quality of life for individuals affected by stroke or spinal cord injury. High-School students as well as undergraduate and graduate students will be trained on this multi-disciplinary project.This research involves learning human intent from biological signals, extracting higher-level goals using sensors embodied in the patient, and developing controllers for motor manipulation based on estimated motor movement intent and higher-level goals. Specific sub-goals proposed to achieve the overall goal of the project include: a collaborative brain-machine learning system that trains the human brain to remap limb movement control to heterologous muscles while simultaneously training the machine to interpret the movement intent from surface electromyograms of the heterologous muscles; algorithms to extract higher-level movement goals using biologic and auxiliary sensor signals; shared brain-machine controllers of robotic hands using the extracted goal and decoded movement intent; and experimental assessment of the capabilities of the methods on individuals with paralysis of the upper limbs. In addition to the innovations in the development of motor prostheses for people with paralysis of the limbs, the proposed research will provide new insights into online learning in nonlinear and time-varying environments, collaborative brain-machine learning, and shared brain-machine control algorithms for motor prostheses.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.