Walking Mode Classification From Myoelectric and Inertial Fusion

Transtibial amputees are generally restricted to the use of non responsive prosthetic systems which have been shown to require increased metabolic energy during normal gait [1] and can contribute to pain in the residual limb [2] among other unfavorable outcomes. Development of responsive lower limb prostheses is restricted due to the function of the human ankle which changes based on speed and type of locomotion. The ideal prosthetic would detect the patient’s motion intent to match both intensity and type of locomotion task. Implementation of motor intent detection should help restore normal limb function. Recent advances have shown the benefit of myoelectric and mechanical sensor fusion towards motion intent classifications. For upper limb amputees, a single accelerometer has shown benefit in classifying different types of hand grasps [3]. For transfemoral amputees a 6 axis pylon-implanted load cell has allowed increased walking mode classification accuracy [4]. It is believed that the continued exploration of EMG and mechanical fusion strategies will advance myoelectric control towards the development of commercially available systems for lower limb amputees. The purpose of the current study was to evaluate the potential for walking mode classification from both electromyography (EMG) signals and inertial measurements units (IMUs).