Human-Robot Collaborative Strategy Based on behavioral intention recognition and switching impedance control for Manipulator

This paper targets small collaborative desktop robots for human-robot handling tasks. It integrates visual-tactile info fusion and motion-intention-based variable impedance control, enabling robotic arm coordination with human operators. A collaborative operation system is built, using object detection/tracking with LSTM networks to predict hand motion intentions. Weight factors defined by segmented functions are dynamically adjusted for smooth transitions and system stability ensured by Lyapunov functions. A fuzzy variable damping strategy based on collaborator motion intention is designed for real-time robotic arm compliance control. Additionally, a variable stiffness strategy allows real-time impedance parameter adjustment. The switching control strategy's effectiveness and smoothness are verified via experiments on a ZED2-UR3e platform, with comparative tests showing the new algorithm's superiority in boosting handling efficiency and user experience.