Functional electrical stimulation (FES) is a neuroprosthesis technique used to restore the motor function of individuals with spinal cord injuries (SCI). For SCI patients, there are some muscles below the injury level which are still innervated, though not volitionally controllable. The principle of FES is to use surface or implantable electrodes to generate pulses of current in intact motor neurons. This is done to induce contraction of these muscles and corresponding joint movement.
Several challenges hinder the application of closed-loop FES outside of research labs, such as that muscles present highly nonlinear and time-varying characteristics. Furthermore, a stimulated muscle changes when fatigue occurs, and muscle models are different for each individual type of muscle. Even more challenging is the fact that there is a significant delay between stimulation and muscle contraction, adding to the processing and transmission delays in the electrical stimulation system.
Research efforts in this project will focus on the development a Model Reference Adaptive Control approach which utilizes the approximation capabilities of Neural Networks. This approach addresses several open problems including uncertain and unmodeled dynamics, actuator dynamics, actuator amplitude and rate saturations, delays, and discrete and real-time implementation. The control algorithm is tested using a muscle-driven forward dynamic model of the lower limb as implemented using OpenSim, a publicly-available musculoskeletal modeling and simulation software environment. An experimental setup is developed to test, understand, and compare muscle dynamics in both open and closed-loop situations.
The merit of this effort includes the extension of current nonlinear control techniques, such as backstepping, dynamic surfacing, and model reference adaptive control, to account for time delays, actuator amplitude and rate saturation limitations, and partial and noisy measurements, thereby substantially increasing the practical applicability of such algorithms. The real-time implementation and the requirement that the FES equipment is easy to setup and simple to use by therapists and patients add additional constraints to the control structure, which needs to be robust yet not overly complicated. Advanced control techniques developed by control engineers have not previously been merged with the advanced neuromusculoskeletal models and the biological understanding of clinicians and biomechanists. Doing so will increase knowledge of muscle characteristics and could lead to the development of an enhanced prosthetic system.
This research effort offers many potential benefits to society, including the possibility of improving the quality of life for patients with paralysis, as well as individuals with other neuromuscular disability including traumatic brain injury, multiple sclerosis, and cerebral palsy. Development of a robust control strategy in cooperation with muscle-driven simulations of movement will provide a framework for guiding rehabilitation strategies for specific impairments. Through a potential future collaboration with rehabilitation researchers and clinicians on the Wake Forest medical campus, the techniques developed as part of this effort will be applied to specific patient populations. From an educational point of view, students from the Mechanical Engineering and Human Nutrition, Foods and Exercise Departments, and Biomedical Engineering will work together to learn and experience in lectures and labs the application of closed loop control techniques to bioengineering problems, bolstering their interest in this field. The interdisciplinary aspects of bioengineering will be covered and disseminated through course development and outreach.
