Research and Implementation of Bionic Manipulator Control System Based on Brain-machine Interface

The aim of this paper is to study and implement a brain-computer interface (BCI)-based control system for a bionic manipulator, which achieves non-intrusive control of the manipulator by combining the modules of EEG signal acquisition, signal pre-processing, feature extraction, pattern recognition and classification, control command generation, and manipulator actuator. The feature extraction stage focuses on frequency and time domain features, and two classifiers, SVM and ANN, are used for pattern recognition and classification. Experimental results show that the SVM classifier achieves an average accuracy of 89% compared to 85% for the ANN classifier, with response times all within 1 second. User feedback showed that most participants were satisfied with the operating experience of the system and were looking forward to its future application prospects. Although there are some problems, such as the discomfort caused by wearing the electrode caps for a long time and the effect of individual differences, these problems have been solved to some extent through measures such as optimising the electrode layout, introducing more training data and providing detailed training guidelines. Future work will focus on improving comfort, enhancing robustness, personalisation, extended functionality, clinical applications and user interface optimisation to further enhance the performance and usefulness of the system.