Abstract: Non-invasive brain-computer interface（BCI） technology, as an emerging human-computer interaction method, has demonstrated broad application prospects in the field of robot control. This study firstly outlines the background and importance of its development, and deeply discusses the physiological basis of brain electrical activity, clarifying how electroencephalography（EEG） has become a common measurement tool for BCI systems due to its non-invasiveness and convenience. Subsequently, this study analyzes the advantages and disadvantages of typical EEG paradigms and applicable scenarios-including active ones such as motor imagery, reactive ones such as steady-state visual evoked potential（SSVEP）, event-related potential P300, and hybrid paradigms that combine the advantages of multiple paradigms. hybrid paradigms that combine the advantages of multiple paradigms, showing how these paradigms can realize complex and efficient robot control tasks. In addition, this study systematically introduces the key steps from EEG signal acquisition to preprocessing and pattern recognition, emphasizes the role of deep learning in improving decoding accuracy, and also points out its challenges, such as high data volume requirements and poor model interpretability. Finally, this study summarizes the development trends and research challenges of BCI technology, and proposes directions to promote the further development of non-invasive BCI technology in practical robot control applications. In summary, this study not only provides an exploration of the application of non-invasive BCI technology in robot control, but also emphasizes the transformative impact that this technology may bring in the future, providing reference and inspiration for subsequent research.

Key words: brain-computer interface, robotic control, motor imagery, steady-state visual evoked potential, pattern recognition