A synchronous robot control system based on the sEMG signals of human upper limb motions

In recent years, the human-computer interaction (HCI) systems (especially the exoskeletal rehabilitation systems) based on surface electromyography (sEMG) signals have been an important research direction. As previous researches indicated, some disabled individuals are unable to contract their muscles by themselves, but their sEMG signals do exist. If we can develop a follow-up control system based on sEMG signals, it is possible to help the disabled individuals with their rehabilitation. In this paper, we proposed and realized a synchronous robot control system by recognizing the sEMG signals of human simple upper limb motions. Specifically, we designed a follow-up control system based on three-class motions. Furthermore, we extracted the feature of mean absolute value (M AV), and applied threshold-based method to train an offline classifier used to recognize the online sEMG signals. The average offline classification accuracy reached 95.7% and the best one reached 98.3%. The average online control success rate reached 88.8% and the best one reached 95.2%. The experimental results indicated that the proposed control system could be considered as the idealized model of wearable exoskeleton system and the approach of this study contributed to realize the clinical rehabilitation system for disabled individuals.