Towards zero training for myoelectric control based on a wearable wireless sEMG armband

The long and complicated preparing procedures, including placement of surface electromyography (sEMG) sensors and intensive training phase before the usage of myoelectric control interface, are key factors inhibiting the extensive applications of sEMG based human-machine interface (HMI). This paper presents an 8-channel sEMG signal acquisition armband (called SJT-iMyo), with the aim of integrating and miniaturizing the sEMG acquisition system. Taking advantage of armband-shape design and Bluetooth wireless interaction, SJT-iMyo is made portable and wearable. Testing results show that SJT-iMyo is capable of acquiring sEMG signals similar to commercial systems. Furthermore, based on SJT-iMyo, a group training framework (GTF) is proposed for the purpose of developing a multi-user myoelectric interface, which is able to be used by new users without training. The proposed method allows pre-training the control interface by a group of users' data, and subsequently testing it by new users without any training or calibration, obtaining 85% classification accuracy and outstanding real-time control performance for 7 motions. This armband can provide effective information to decode movement intent of users, and overcome individual differences combining with the GTF. The promising outcomes of this study have the potential for promoting the practical applications of sEMG based HMI.