Efficient upper limb joint displacement modeling using EMG signal for driving an assistive SCARA

Abstract

Within the scope of wearable robotics, human to machine interfacing is a critical element of the development of an accurate, reliable, and apt device. Wearable robotics span through a varied field of rehabilitation robotics, exoskeletons designed for industrial, military and personal use, and recently, even for handicap offset. Within any of these varied uses, a key design step is defining a viable, consistent source of user intent. Without a clear vision of what the user wishes to accomplish, the robotic aid cannot behave in a way that is of any benefit to the user in question. The use of electromyography (EMG) is widely regarded as a viable means of correlating signals obtained from the body to physiological locomotion. The non-invasive nature of surface EMG serves as a comfortable method of extracting accurate musculoskeletal functions for this purpose. We purpose to accurately determine user intent through the processing and feature extraction of biological signal sources. The current work will focus on the development and methodology used to correlate joint angles with EMG signals for a planar model used in a developing neurological rehabilitation robot based on a Denso HS-4555 Selective Compliance Articulated Robot Arm (SCARA).