A group brain-controlled method for UAVs using a hybrid paradigm of hand movements and visual evoked potentials

Brain-computer interfaces (BCIs) are among the most prominent communication technologies that establish a direct channel for information exchange between the brain and external devices. They have been extensively applied in the field of aerospace. However, traditional BCI technology faces challenges, including a limited number of brain control commands and insufficient recognition accuracy in electroencephalography (EEG) decoding. These limitations make it difficult for traditional BCIs to perform complex tasks with high accuracy. Therefore, this study proposed a novel group BCI (G-BCI) system and further constructed a brain-machine shared control method for unmanned aerial vehicle (UAV) swarm control. First, a novel G-BCI paradigm combining precise hand movements and visual evoked potentials was designed. Moreover, an improved multi-domain feature fusion convolutional neural network (MDFF-CNN) was employed to decode EEG and electromyography (EMG) signals from precise hand movements, while a Filter Bank Common Spatial Patterns with Canonical Correlation Analysis (FBCCA) method was used for Steady-State Visual Evoked Potentials (SSVEP) decoding. Furthermore, a task-driven shared control model mapping the G-BCI system and the leader-follower UAV swarm control strategy was proposed. To verify the effectiveness of the proposed method, eight participants were recruited to conduct both offline and online experiments. The proposed G-BCI system achieved an offline accuracy of 88.91 ± 5.06% and an online accuracy of 88.89 ± 1.96%. All the experimental results demonstrate the feasibility of the proposed method.