Compliance discrimination via biomimetic optical tactile sensors and human-like impedance regulation

Endowing robots with biomimetic tactile sensors enhances contact information retrieval. In previous work, we proposed a vision-based approach inspired by human tactile perception to discriminate object compliance with a soft biomimetic tactile optical sensor (TacTip). In this work, we first enhanced the vision-based method by improving the accuracy of the estimation of the initial contact area. Then, we integrated the mechanism of internal muscular regulation (co-contraction) that humans adopt to maximize the information uptake during object compliance probing. To this aim, we used human co-contraction patterns extracted during object softness probing to control a Variable Stiffness Actuator used to actuate the indentation system endowed with the TacTip. The obtained results show that the proposed improvements significantly enhance our model-based approach.