Abstract :
Sensing surface textures by touch is a valuable capability for robots. Until recently it was difficult to build a compliant sensor with high sensitivity and high resolution. The GelSight sensor is compliant and offers sensitivity and resolution exceeding that of the human fingertips. This opens the possibility of measuring and recognizing highly detailed surface textures. The GelSight sensor, when pressed against a surface, delivers a height map. This can be treated as an image, and processed using the tools of visual texture analysis. We have devised a simple yet effective texture recognition system based on local binary patterns, and enhanced it by the use of a multi-scale pyramid and a Hellinger distance metric. We built a database with 40 classes of tactile textures using materials such as fabric, wood, and sandpaper. Our system can correctly categorize materials from this database with high accuracy. This suggests that the GelSight sensor can be useful for material recognition by robots.
Keywords :
image recognition; image sensors; image texture; robot vision; GelSight sensor; Hellinger distance metric; compliant sensor; fabric; height map; local binary patterns; material recognition; multiscale pyramid; robots; sandpaper; surface texture recognition; surface texture sensing; tactile textures; visual texture analysis; wood; Databases; Histograms; Microstructure; Robot sensing systems; Surface texture; Surface treatment; local binary pattern; tactile sensing; texture recognition;
