Quantifying the natural tactile scene

An animal’s only access to the environment are the sensory signals provided by the animal’s sensors in response to its interaction with the external world. Thereby, the information conveyed by these sensory signals are merely implicit and relate to the physical quantities in the environment in a highly nonlinear manner. Nonetheless, nervous systems are able to infer the environmental variables based on the sensory input signals they evoke with extraordinary efficiency and reliability. It was hypothesized that this is mainly achieved by exploiting regularities and redundancies in the environment and sensory input signals. A large body of research, also known as natural scene statistics, has generated strong evidence confirming this idea in vision and audition. The goal of this project is to quantify the natural scene for tactile systems using curvature information from 3D scans of natural environment.