A Hybrid Computer Interface for Robot Arm Control

Abstract

Brain-computer interface (BCI) directly translate human thought into machine command. It provides a new and promising method for rehabilitation of persons with disabilities. BCI actuated robotic arm is an effective rehabilitation way for patients with upper limb disability. Based on the study and reference of the existing brain-controlled robot arm, this paper proposed a method of combining electromyography (EMG) and Electroencephalogram (EEG) to control the manipulator. Specifically, we collect EMG signals from the human leg and use the leg movements to quickly and reliably select the joints which are currently activated. The robot arm joints are precisely controlled by movement imagination (MI) brain-computer interfaces. The use of two non-homologous signals, scattered the burden of the brain and therefore reduce the work load. In addition, the program allows two kinds of operations at the same time, so the program is flexible and efficient. Offline experiment was designed to construct the classifier and optimal parameters. In the online experiment, subjects were instructed to control the robot arm to move an object from one location to another. Three subjects participated in the experiment, the accuracy rates of classifiers in the offline experiment were exceeded 95% and they all completed the online control.