Abstract: As a variant of the RSVP paradigm, Dual RSVP exhibits superior classification performance compared to the traditional Single RSVP, showing considerable potential for the development of brain-computer interface control systems. However, current research on the Dual RSVP paradigm remains largely limited to cognitive experiments and has not been further applied to human-computer interaction systems. In light of this, a novel encoding method based on Dual RSVP was designed and an integrated hybrid BCI virtual robotic arm control system was developed based on Dual RSVP paradigm. To verify the effectiveness of the system, 18 college students were recruited to participate in RSVP experiments under both single-window and dual-window paradigms, with synchronous collection of electroencephalogram (EEG) and eye movement (EM) data. The brain-eye fusion classification results of the two paradigms demonstrate that the dual-modal fusion performance of EEG and eye-tracking outperforms those of either EEG or eye-tracking modalities. Moreover, Dual RSVP achieves a higher information transfer rate and classification accuracy than Single RSVP. These results provide valuable references for the development of hybrid brain-computer interface control systems based on EEG and eye-tracking modalities.