Feasibility study of a novel rehabilitation training system for upper limb based on emotional control

This paper introduces a new way to help people who lose motor function regain their abilities of activities of daily living (ADL). As is proved, repeated training can help the paralytics rebuild the strength of muscles, while active participation of the patients will improve the outcome of rehabilitation than them barely assisted by robots. Brain-computer interface (BCI) technology brings paralytics a new way to join in the training process actively. The novel idea of our study is to treat the robot as a co-worker and the paralytic interacts in the training process with emotional states based on his satisfaction level of the robot's work. This system consists of two main parts: the robot and BCI. The robot can work in three different modes in training process. This paper focuses on the feasibility study of emotional control of the robot. Experiments were conducted with a healthy male subject. Brain signals of the subject were extracted by an electroencephalogram (EEG) headset. Four different mental states were detected and interpreted into control commands to start the robot, stop the robot, maintain training mode and switch training mode. Raw EEG data were recorded for off-line analysis. Powers of four frequency bands of EEG signals were analyzed to find their relationship with different emotional states. Experimental results proved the feasibility of our system and shown that it's much easier to control the training process with the collaborating of the semi-autonomous robot. We also found that different EEG frequency bands carry important messages about different emotions, which may provide references for further study of emotional control.