Automatic adaptive onset detection using an electromyogram with individual difference for control of a meal assistance robot

Abstract

The onset detection of muscle activation is an essential issue in electromyogram (EMG) control. In this paper, a novel approach based on EMG power with automatic adaptive threshold is proposed to address this issue. The purpose is to develop an effective EMG-controlled meal assistance robot. Taking into account the individual difference such as contraction power and resting power, the threshold of onset detection is set with respect to the latest EMG signal. The results show the method is able to adjust automatically to avoid false alarms, and works well when the contraction power varies. Implementation of this EMG-controlled meal assistance robot may provide limb-deficient patients with an effective and comfortable human–machine assistance interface.

Keywords:

• Electromyogram
• Automatic adaptive onset detection
• Contraction power
• Individual difference
• Meal assistance system