Brain-actuated Humanoid Robot based on Brain-computer Interface (BCI)

Abstract

Brain-computer interface (BCI) technology provides a new communication and control channel between the brain and a computer, without the participation of peripheral nerves and muscles. This paper proposed a brain-actuated BCI paradigm to control a humanoid robot. Because of the extremely low signal-to-noise ratio (SNR) of the electroencephalograph (EEG) signals, effective BCI paradigm for multiple degrees of freedom (DOF) control without external stimulation is hard to construct. To address this problem, a novel sequential coding method was presented in this paper to increase the number of output commands. With this method, different brain activities could be modulated using one motor imagery task only. A four-class BCI paradigm was designed based on the sequential coding method, and used to control a humanoid robot. Four subjects participate in the walking experiment of the robot. The average time cost to complete one walking task was 139.8 seconds. The experimental results demonstrated the effectiveness of the sequential coding method for multi-class BCI paradigm design.