Intention-aware hybrid motion-force control with weighted normal estimation for telerobotic ultrasonic testing

Telerobotic ultrasonic testing is a promising approach for equipment maintenance in hazardous and complex environments, as it enables human operators to remotely control robots to perform defect detection tasks. However, unmodeled disturbances between the probe and the workpiece surface during the contact can degrade the control accuracy of the system, while inevitable normal alignment errors caused by unknown surface geometry can further reduce the quality of defect detection. To address these issues, a hybrid motion-force control framework integrating dynamic model compensation and state feedback control is proposed to enhance system robustness against unmodeled disturbances while ensuring precise control of the robot’s position, orientation, and contact force. Secondly, a coarse-to-fine desired direction tuning model for the robot end-effector is proposed, which incorporates the human intention to improve the accuracy of normal alignment under inevitable normal estimation errors and to enhance the system adaptability to dynamic tasks. Finally, the accuracy and advantages of the proposed method are validated through ablation studies in simulation and real telerobotic ultrasonic testing experiments on a curved workpiece. In real experiments, it reduces the position tracking error by 81.7% and the orientation tracking error by 39.7%, and improves the quality of the detected ultrasonic waveform by 51.1%.