DIBLF-Based Adaptive Optimal Constrained Control for Collaborative Robots Under Different Human-Robot Interactive Tasks

The flexibility and safety of physical human-robot interaction are essential for real-world applications. Therefore, this study investigates adaptive optimal control for physical human-robot interaction under dynamic output constraints. We develop an admittance-based approach to reconstruct reference trajectories, facilitating smooth online transitions between different interactive tasks. Additionally, we introduce a regulation function that establishes the relationship between interaction force and various collaborative robot behaviors. To accommodate more general dynamic output constraints, we propose a dynamic integral barrier Lyapunov function (DIBLF)-based adaptive dynamic programming control scheme, which extends the applicability of the integral barrier Lyapunov function (IBLF) to a wider range of cases. Stability analysis shows that all signals in the closed-loop system remain bounded, and the output constraints are consistently upheld. Finally, a Franka EMIKA Panda robot is utilized as a test platform to perform a material deposition task, thereby validating the effectiveness of the proposed methodology.