In vivo and in vitro comparative assessment of the log-linearized Hunt–Crossley model for impact-contact modeling in physical human–robot interactions

Abstract

In physical human–robot interactions, making the robot perceive in real time the mechanical contact impedance is critical for interactions safety, robot control and haptic rendering for robot teleoperation and can be achieved through online parametric model identification. Probing the viscoelastic properties of tissues is also a medical concern. For soft viscoelastic biological tissues, the Hunt–Crossley model is a contact force model computationally inexpensive while being accurate. As this model is non-linear, a log linear approximation has been proposed to achieve a fast and real-time identification using a recursive least squares approach. In this article, we want to regard the log-linearized expression of the Hunt–Crossley model no more as an approximation, but rather as a valuable empirical mechanical model of soft biological tissues. We show through experimental data fit and sophisticated statistical analysis that the log-linearized Hunt–Crossley model performs always closely to the Hunt–Crossley model and is even often slightly better. The experimental conditions investigated are related to impact and contact interactions, relevant in the context of Cobotics.