Real-time myoelectric control for a lower-limb assistive exoskeleton

This paper describes the approach to a novel algorithm for real-time myoelectric onset detection based on the classic moving average method with added features to provide onset detection while myoelectric signals are recorded from two muscles at a time. The aim of this algorithm is focused on the generation of real-time binary signals to drive a dc motor attached to the hip joint of a rehabilitation exoskeleton for the assistance on lower-limb rehabilitation of hemiplegic patients in supine/prone, seated and standing positions. With this low computational cost real-time control algorithm based on myoelectric signals, patients involved on a physical rehabilitation program wearing the developed exoskeleton will have full control over the assistance for the lower-limb exercises. The parameter tuning for the myoelectric onset detection algorithm and the results of its accuracy along several tests over two test subjects on the rectus femoris and semitendinosus muscles are presented.