A Complaisant and Impact-Resistant Rigid–Flexible Controller for Physical Human–Robot Interaction Considering the Possible Interferences

Abstract

In physical human–robot interaction (pHRI), guiding forces combined with impacts often cause the robot to deviate from the intended human trajectory. To address this issue, a finite-time neural network terminal sliding mode adaptive robust integration controller and a position-based adaptive admittance mapping model are proposed. The proposed trajectory tracking controller exhibits fast convergence in finite time, while also demonstrating robustness and precise trajectory tracking performance. This establishes the foundation for tracking the reference trajectory generated by the admittance model. Subsequently, the adaptive admittance mapping model decouples dynamic forces, adjusts system damping, and generates the reference trajectory, enabling the trajectory tracking controller to compute the corresponding control torques. This allows the robot manipulator to exhibit compliance during pHRI tasks, while maintaining stability under sudden impacts. Extensive experiments were conducted, demonstrating the precision, robustness, and rapid convergence of the proposed trajectory tracking controller. Additionally, an impact resistance experiment was conducted, confirming the impact resistance capability of the rigid–flexible controller.