This project will develop a Hierarchical Hybrid Controller (HHC) for a Neural Prosthesis (NP) that uses Functional Electrical Stimulation (FES) for assisting walking of individuals who suffered a central nervous system injury. The HHC would mimic biological control by using a rule-base model of walking at the higher coordination level, and a customized model-based controller at the lower actuator level. This approach was selected based on neuroscience studies and includes both descending and reflex components of the control of walking. The project will address issues such as how to synthesize rules by using machine learning; the methods needed to incorporate individual characteristics are instrumental for the actuator level of the control. The machine learning would consider adaptive logic networks and radial-basis function neural networks. The model-based control (actuator level) will rely on the reduced spatial models of bipedal walking. The controller also considers the following important constraints: high safety, high reliability, ease of donning and doffing, and user interface at the subconscious level. The use of this controller is envisioned in NP with surface electrodes during neurological rehabilitation, as well as implantable NP as orthotic devices. Individuals after stroke, incomplete spinal cord injury or multiple sclerosis who can stand and ambulate with minimal or no arm support are likely the ones who will benefit from this new control strategy. (Mirjana Popović, Thomas Sinkjær, Dejan Popović, Erika Spaich)
Effektiv start/slut dato01/01/200630/06/2008
ID: 214519616