Technology and Disability 15 (2003) 95–103 95 IOS Press Continuous multifunction myoelectric control using pattern recognition Kevin Englehart a,* , Bernard Hudgins a and Adrian D.C. Chan b a Institute of Biomedical Engineering, University of New Brunswick, Fredericton, NB, Canada b Department of Systems and Computer Engineering, Carleton University, Ottawa, ON, Canada Abstract. This work represents an ongoing investigation of dexterous and natural control of upper extremity prostheses using the myoelectric signal. The scheme described within uses pattern recognition to process four channels of myoelectric signal, with the task of discriminating six classes of limb movement. The method does not require segmentation of the myoelectric signal data, allowing a continuous stream of class decisions to be delivered to a prosthetic device. Due to the fact that the classifier learns the muscle activation patterns for each desired class for each individual, a natural control actuation results. The continuous decision stream allows complex sequences of manipulation involving multiple joints to be performed without interruption. The continuous classifier is optimized with respect to the feature set and classifier used, and post-processing of the decisions to eliminate spurious errors. Keywords: EMG, myoelectric, pattern recognition, prostheses, classification 1. Introduction The surface myoelectric signal is an effective and important system input for the control of powered pros- theses. This control approach, referred to as myoelec- tric control, has found widespread use for individuals with amputations or congenitally deficient upper limbs. Clinical evaluations of myoelectrically controlled pros- theses indicate that the three major factors which deter- mine the acceptance rates by the users are: the type of prosthesis, the degree of user training, and the control strategy. It is the third factor that we consider here. It has been observed [2] that low acceptance rates result when the user perceives an inadequate controllability – specifically a lack of intuitive and dexterous control. A myoelectric control system that offers exceptional per- formance is described with regard to three important aspects of controllability: the accuracy of movement * Address for correspondence: Kevin Englehart, Ph.D., P. Eng., Associate Professor, Electrical and Computer Engineering, Associate Director, Institute of Biomedical Engineering, University of New Brunswick, Fredericton, NB, Canada E3B 5A3. E-mail: kengleha@ unb.ca. selection, the intuitiveness of actuating control, and the response time of the control system. – Accuracy is essential to faithful realization of a user’s intent. Accuracy must be as high as possi- ble, although it is difficult to define the threshold of acceptability, as no definitive clinical trials have addressed this issue. – An intuitive interface to the control system relieves the mental burden of the user. In this regard, a control system should be capable of “learning” the muscle activation patterns chosen as the most “natural” by an individual to actuate motion. – The response time of a control system should not introduce a delay that is perceivable by the user. This threshold is generally regarded to be roughly 300 ms. This places a real-time constraint on the control system’s tasks of acquiring and processing myoelectric data. 2. Background The concept of myoelectric control was introduced in the 1940’s [28], however the technology of the day ISSN 1055-4181/03/$8.00  2003 – IOS Press. All rights reserved