The Departments of Industrial & Systems Engineering (ISE) and Computer & Information Science & Engineering (CISE) at the University of Florida received a grant from the National Science Foundation to develop machine learning algorithms to drive prosthetic devices to better assist amputees in their daily lives.
Limb amputation is a major cause of disability. Over 100,000 people in the United States have upper limb amputations. Many of these cases occur at the forearm or upper arm, causing victims to lose the ability to complete everyday tasks such as brushing their teeth or dressing themselves. With the help of modern technology and the use of prosthetic limbs, patients can regain some of these functions. However, prosthetic devices and interfaces are often costly and difficult to learn, resulting in frustration and abandonment of the device altogether.
David Kaber, Ph.D., Chair of the Department of Industrial & Systems Engineering at UF, and He Huang, Ph.D., a professor of biomedical engineering at North Carolina State University, are working together to develop better technology to make these prosthetic devices more accessible and user-friendly for amputee patients.
“This research is very human-oriented, and it is ultimately intended to promote a better quality of life for patients that have experienced traumatic injuries. The technology has to be better in order to ensure these devices are more user-friendly and cost-effective so that they can be utilized by patients,” said Kaber.
Huang, who has an expertise in working with powered prosthetics, has designed an interface, much like a patch, that is dense with motion sensors that detect the electrical activity of the remaining muscle at the radius or humerus level, and then translate that movement to control the device. The interface captures the pattern of electrical activation at the residual muscle, which is generated by a patient’s conscious thought of a specific upper limb movement, such as opening or closing the hand. This pattern of activity is then classified and associated with activation of servo motors as part of a powered prosthetic to cause opening/closing of a mechanical gripper. The same muscle activation pattern classification, and association with other prosthetic functions, is completed for a range of upper limb movements.
Once one of these devices is prototyped, patients must go through extensive training in order for the neural network to capture each individuals’ patterns of electrical stimulation of muscle for commanding the prosthetic. The training of the device is customized to the individual user.
Kaber and the project research team have posited that this electromyography interface could be used to control other devices and technologies in addition to powered prosthetics, such as computers. He and Jaime Ruiz, Ph.D., UF CISE assistant professor, are working to demonstrate the use of virtual reality as a tool to assess the electromyography by requiring patients to manipulated virtual objects, as opposed to prosthetic devices, while immersed in the virtual task environment.
Once patients are able to effectively use this technology to control a powered prosthetic or control a computer, the last phase of the project will test the interface’s capability for supporting control of real systems, such as a motor vehicle. This research will be conducted in the Human Systems Engineering Laboratory at the University of Florida using the lab’s new full-scale and full-motion driving simulator, which will test a patient’s ability to cognitively think about controlling certain vehicle functions like steering and braking.
The goal of this research is to provide greater accessibility for amputee patients to effectively and affordably engage in everyday activities.
The three-phase project is a collaboration with North Carolina State University and Texas A&M and will last for three years.