Integrating Human Intention into Multirobot Decision Making via Brain–Computer Interface Enabled Shared Autonomy

Multirobot systems tend to have higher execution efficiency when performing tasks such as mapping, search, and space exploration. Still, due to the influence of sensor measurement error, the decision-making of multirobot systems usually has deviations that are difficult to eliminate. In this unfavorable situation, the advantage of human experience can generally guide the multirobot system in making the correct decision. This study proposed a high-resolution brain–computer interface (BCI) paradigm and constructed a human intention probability model through graph neural networks. This allows for the preservation of richer interactive information, capturing the inherent uncertainty and preference features of human intention. Meanwhile, a BCI-enabled shared autonomy strategy integrating probabilistic human intention through opinion dynamics is introduced, ensuring the collaborative participation of humans and robots in decision-making. Experimental results show that the shared autonomy approach significantly improves decision-making accuracy compared to the initial multirobot estimate. Further analysis shows that this approach greatly outperforms traditional BCI strategies, showing promise for human–multirobot cooperation in complex task environments.