Abstract of Prof. Etienne Burdet's Talk

Talk title: Haptic invariants in humans, for robots.

Abstract: 

When searching for an object in my pocket, I can clearly perceive the edges, shape, and hardness of the items I touch—regardless of which part of my fingers makes contact and despite large sensorimotor noise. What accounts for these perceptual invariants? Key elements to investigate this question are revealed by analyzing mechanical spatiotemporal invariants emerging from the constraints of skin–body–environment interactions, alongside corresponding neural invariants that could be leveraged to create haptic perception in prosthetics. 

We follow a similar approach to create haptic perception in robotics. I will present a modular electronic skin designed to exploit the intrinsic mechanical properties of compliant sensing, enabling the emergence of informative invariant features. I will also show how such haptic invariants can be extracted through machine learning to support robust robotic perception across varying contact conditions. Overall, this perspective suggests shifting from raw signals to physically grounded invariants, opening new directions for robotics and next-generation prosthetics.