Evaluating Populations of Tactile Sensors for Curvature Discrimination.

Abstract

The high density of receptors in fingertip skin is a limiting factor for replicating tactile feedback for neural prosthetics. At present, the large size of engineered sensors and the dense network of neural connections from finger to brain inhibit duplicating the approximately 100 receptors/cm(2). The objective of this work is to build a model of the skin and neural response with which populations of sensors can be positioned and evaluated when discriminating spheres. The effort combines a 3D finite element model of the fingertip, a bi-phasic transduction model, and a leaky-integrate-and-fire neuronal model. Populations of sensors are configured with three average densities (10,000/cm(2), 1,000/cm(2), and 100/cm(2)). For these populations, the firing rates for the dynamic (40-70 ms) and static (650 ms-900 ms) phases and first spike latencies are predicted. The model can differentiate indenters at a level similar to human performance at each sampling density, including of the human finger (100/cm(2)).