Abstract: A hybrid brain-computer interface (BCI) system based on SSVEP and attention EEG is designed, for the problems of multi-classification for attention EEG and the expansion of the experimental paradigm of the BCI system. Firstly, the multi-classification of attention EEG is explored, and an attentional paradigm that integrates spatial attention and mental tasks is designed to improve the classification performance of attention EEG. In the brain-controlled robot experiment, the α-block signal is used to switch SSVEP and attentional stimulation interface. SSVEP signal is used to control the movement of the robot. Attention EEG is used to change the speed of the robot moves. Ten subjects are recruited to experiment with the brain-computer interface system. The results show that the accuracy of three-classification of attention signals reach 78.2%. The correct rate of the robot reaching the target position can get to 100%.