摘要：AbstractAlthough many hand exoskeletons have been developed to provide natural hand movements for patients, extraction of motion intents is one of the challenging issues to perform activities of daily living (ADLs). This paper describes a noninvasive electromyography (EMG)-based proportional control for grasping tasks. To perform grasping tasks in ADLs, three degree of freedoms (DOFs) for thumb adduction/abduction, thumb flexion/extension, and four fingers flexion/extension are required at least. In this work, a 2-D Fitts’ Law test is designed as a suitable alternative of real grasping tasks to perform a real-time EMG control for three DOFs required for grasping. The test setup was built to evaluate a user can generate three commands for direct proportional EMG control. Three EMG signals induced by voluntary wrist and thumb movements were recorded to extract users’ intentions, and each signal was used to activate each degree of freedom in Fitts’ law task. A user was asked to move the cursor into the target using three commands such as cursor right, cursor up, and cursor to the initial position. Healthy subjects were performed our developed interface and the results show that the developed interface performed better than that of our previous study using a pattern recognition method. Although there are several issues remain to be resolved, we expect that proportional EMG control could be used to help SCI or stroke patient for controlling a hand exoskeleton.
