Tactile sensing using elastomeric sensors

Abstract

GelSight, namely, elastomeric sensor, is a novel tactile sensor to get the 3D information of contacting surfaces. Using GelSight, some tactile properties, such as softness and roughness, could be gained through image processing techniques. In this thesis, I implemented GelSight principle to reconstruct surface geometry of tested surfaces, based on which, the roughness comparison and lump detection experiment are conducted. Roughness of five different types of sandpapers are successfully compared using GelSight Ra value. In the lump detection experiment, a visual display for tactile information is presented. To get binary feedback of lump presence or not, a simple threshold method is introduced in this thesis. To evaluate the performance of GelSight sensor, human psychological experiments are conducted. In similar tasks, GelSight sensor outperforms humans in lump detection.