Perceptual model for compliance in ineraction with compliant objects with rigid surface

In the design and control of robotic sensor system or haptic rendering system, one key technology is to give robotics the ability of feeling compliance like human. An easy way to realize it is quantitatively describing the relationship between stimuli and subjective perceptual results. In this work, we address the problem of mapping the interactive data of compliant objects to the perceptual compliance using perceptual model. Firstly, a perceptual space of compliant objects was built though psychophysical experiments and the N-MDS method. Then effective features were selected to estimate human-perceived differences of compliance based on correlation analyses. Finally, a perceptual model was built based on regression analysis to describe the subjective perception using effective feature parameters. Results show that a one-dimensional embedding space of “hardness” is proper to describe the subjective perception. Features of global stiffness and ratio of force and displacement could fit the perceptual results well in the form of exponential expression.