Regulating human adaptive impedance is essential for effective physical human–robot interaction (pHRI). Traditional stiffness estimation methods based on electromyography (EMG) are often limited in practical applications due to factors such as skin condition and electromagnetic interference. To address these issues, this paper aims to propose an online method to estimate arm endpoint stiffness using force sensing resistor (FSR) arrays and inertial measurement unit (IMU) sensors. This approach avoids direct skin contact, improving both usability and practicality.
A flexible FSR sensor array is wrapped around the surface of the object being held to measure average palm grip force. IMU sensors record the posture of arm and build a stiffness model of the arm endpoint. The model’s parameters are first identified through a disturbance-based calibration experiment.
The method was tested using a gripping tool mounted on a KUKA robotic platform. Experimental results show that it can accurately estimate arm endpoint stiffness and enable variable impedance control. Compared to traditional sEMG-based methods, the sensor setup time is reduced by 57.1%.
Although EMG is widely used in pHRI for stiffness estimation, its sensitivity and the need for complex skin preparation make it impractical for industrial use. The proposed method addresses these limitations and offers a more practical alternative. It contributes to the advancement of variable stiffness control and human–robot collaboration, with promising applications in intelligent manufacturing and rehabilitation.
