Robust control barrier function-based shared control for cognitive-physical human-robot collaboration

Ensuring safe shared control in human - robot collaboration remains challenging due to uncertain human inputs and time-varying operator cognitive states. Existing methods primarily address either physical-interaction safety or authority allocation, but rarely provide a unified framework that simultaneously enables cognition-aware authority adaptation and formal safety guarantees. To address this issue, this paper proposes a robust coupled cognitive - physical shared-control framework for human - robot collaboration. First, an augmented state-space model is established by integrating robot dynamics with operator cognitive states, where the human control input is explicitly treated as a bounded disturbance. Based on this model, multiple robust control barrier functions are constructed to enforce obstacle avoidance, velocity limits, and lower bounds of cognitive safety levels via an online quadratic-programming-based controller. Furthermore, a cognition-driven dynamic authority allocation mechanism and a hierarchical intervention strategy are introduced to enable adaptive transitions between human-dominant and robot-dominant modes. The proposed framework guarantees forward invariance of the safe set and bounded closed-loop signals. Simulation results under uncertain human input and cognitive degradation scenarios demonstrate improved safety, adaptability, and collaboration compared with conventional methods.