Volitional Control of a Prosthetic Knee Using Surface Electromyography

• Published in: IEEE transactions on biomedical engineering 2011-01, Vol.58 (1), p.144-151
• Main Authors: Ha, Kevin H., Varol, Huseyin Atakan, Goldfarb, Michael
• Format: Article
• Language: English
• Subjects: Amputees; Biological and medical sciences; Diseases of the nervous system; Electrodes; Electrodiagnosis. Electric activity recording; Electromyography; Electromyography - methods; Humans; Impedance; Investigative techniques, diagnostic techniques (general aspects); Joints; Knee; Knee - physiology; Knee Prosthesis; lower-limb prosthesis; Medical sciences; myoelectric control; Nervous system; powered prosthesis; Prosthetics; Quadriceps Muscle - physiology; Radiotherapy. Instrumental treatment. Physiotherapy. Reeducation. Rehabilitation, orthophony, crenotherapy. Diet therapy and various other treatments (general aspects); Range of Motion, Articular; Signal Processing, Computer-Assisted; Trajectory; transfemoral prosthesis
• ISSN: 0018-9294; 1558-2531
• DOI: 10.1109/TBME.2010.2070840

This paper presents a method for providing volitional control of a powered knee prosthesis during nonweight-bearing activity such as sitting. The method utilizes an impedance framework, such that the joint can be programmed with a given stiffness and damping that reflects the nominal impedance properties of an intact joint. Volitional movement of the knee joint is commanded via the stiffness set-point angle of the joint impedance, which is commanded by the user as a function of the measured surface electromyogram (EMG) from the hamstring and quadriceps muscles of the residual limb. Rather than using the respective EMG measurements from these muscles to directly command the flexion or extension set point of the knee, the presented approach utilizes a combination of quadratic discriminant analysis and principal component analysis to align the user's intent to flex or extend the knee joint with the pattern of measured EMG. The approach was implemented on three transfemoral amputees, and their ability to control knee movement was characterized by a set of knee joint trajectory tracking tasks. Each amputee subject also performed the same set of trajectory tracking tasks with his sound side (intact) knee joint. The average root mean square trajectory tracking errors of the prosthetic knee employing the EMG-based volitional control and the intact knee of the three subjects were 6.2° and 5.2°, respectively.