Flexible Multimodal Tactile Sensing System for Object Identification

Abstract

This work presents results towards realizing a flexible multimodal tactile sensing system for object identification. Using polymer substrates and simple fabrication, robust devices are made that can identify objects based on texture, temperature, as well as material properties such as hardness and thermal conductivity. These capabilities are possible using signal processing techniques and physical models along with individual sensing structures inspired by the specialization found in biological skin. These structures are used to sense various object parameters, and array-wide processing to identify texture using a Maximum Likelihood decision rule. 80% texture classification is achieved. In blind object identification tests, over 90% correct identification was achieved by measurement of material properties.