EMG based Pattern Recognition of Human Lower Limb Motion using AR Model and LS-SVM

Tong Mu, Xiaodong Zhang, and Binghui Jia


sEMG, Human lower limb, seamless offload, IP flow mobility


Persons with disabilities, old people or mountain climber have closer relationship with the prosthesis and rehabilitative robot, which offers the power to help walking and further enhance the capacity and speed of motion with heavy load or long-time motion. It has important meaning to accomplish the precise control with prosthesis or rehabilitative robot after accurately classifying the human motion pattern. First, AR-parameter model is discussed for extracting sEMG signal from biceps femoris, rectus femoris, vastus medialis and gastrocnemius with the human motion of running, walking and standing, and then least squares support vector machine (LS-SVM) is analyzed for classifying the motion patterns. The results show the methods of pattern recognition used in this paper have the recognition rate of 83.33%, and it shows a good application prospect on pattern recognition of human lower limb.

Important Links:

Go Back