Local Shape Estimation Using Mechanochromic Structurally-Colored Tactile Sensors

Abstract

Tactile perception is increasingly used in robotics to augment the robot’s sense of its environment and the objects that it manipulates, particularly in cases where visual systems alone prove inadequate. Existing tactile sensors feel their surroundings by sensing many different types of tactile signals such as pressure, contact position, or contact shape. This thesis introduces a novel method for measuring and reconstructing the shape of an object based on a tactile imprint, and describes the framework for this method, the fabrication of the necessary materials, and the subsequent testing and validation of the process. The procedure outlined in this work involves the use of a custom tiled mechanochromic structurally-colored film in conjunction with a digital camera, and calculates shape and strain information of the film based on how the observed colors shift when undergoing deformation. When combined with a transparent elastomeric pad, this arrangement can be used to deduce information about objects that are pressed into the pad by observing the deformation in the surface. This ability to measure the shape and strain state of a surface by leveraging the high resolution of modern image sensors together with color-dynamic films tiled in a checkered pattern may allow for more effective tactile sensors, and more broadly can provide a useful tool for research and industrial applications. While this work focuses specifically on tactile shape reconstruction, the methodology presented can similarly be applied to more general cases where shape or strain information of a surface is desired.