Adaptive Variable Impedance Control in Physical Human–Robot Interaction Based on Arm End Stiffness Estimation

Impedance control is widely utilized in physical human–robot interaction (pHRI), particularly in rehabilitation and medical applications. However, traditional constant impedance control is insufficient for dynamically changing human–robot collaboration tasks, while variable impedance control often underperforms due to the challenge of accurately estimating human stiffness. In this article, an adaptive variable impedance control method based on real-time human stiffness estimation is proposed to significantly extend the applicability of traditional impedance control. First, a stiffness estimation model based on surface electromyography (sEMG) is developed to infer human intention by designing a visual feedback perturbation method for data acquisition and parameter identification. Then, a constraint based on a velocity mapping band is implemented to enhance compliance during interaction, while an adaptive variable stiffness constant force tracking algorithm is introduced to meet the force control requirements of specific task scenarios. Furthermore, two variable impedance control strategies incorporating human intention are proposed for different collaborative interaction scenarios. Finally, the effectiveness of the proposed method is validated through two human–robot collaborative tasks: sawing and wiping.