Abstract: The stiffness information of the grasped object at the initial contact stage can be effectively used to adjust the grasping force of the prosthetic hand, thereby preventing damage to the object. However, the object’s deformation and contact force are often minimal during the initial stage and not easily obtained directly. Additionally, stiffness estimation methods for prosthetic hands often require contact sensors, which can easily lead to poor contact issues. To address the above issues, this paper proposes the model-based stiffness estimation of grasped objects for underactuated prosthetic hands without force sensors. First, the kinematic model is linearized at the contact points to achieve the estimation of the linkage angles in the underactuated prosthetic hand. Secondly, the motor parameters are estimated using the Kalman filter method, and the grasping force is obtained from the dynamic model of the underactuated prosthetic hand. Finally, the contact model of the prosthetic hand grasping an object is established, and an online stiffness estimation method based on the contact model for the grasped object is proposed using the iterative reweighted least squares method. Experimental results show that this method can estimate the stiffness of grasped objects within 250 ms without contact sensors.

Key words: Stiffness estimation')">Stiffness estimation, Noncontact sensor, Underactuated prosthetic hand, Kalman filter, Iterativereweighted least squares, Contact model