Designing a Tactile Sensor Array for Shape Recognition

This paper describes the design of a tactile sensor array that used nine force sensors to identify and categorize the shape of an object. This system included a silicone surface with an artificial silicone fingerprint-like surface. From the measurements, the normal and shear forces were converted into visual pixel images. Cellular Nonlinear Network (CNN) templates were applied to determine the timing and type of spatial-temporal events. In pilot experiments, the CNN templates were able to determine both when a force was applied and the curvature of the plate applying the force.