Speed-invariant Texture Discrimination using an Optical Tactile Sensor Array (LiVec Finger)

Authors:

Abstract
Providing robotic fingers with tactile sensation can enable better performance of dexterous manipulation. Identifying a texture by touch can further the ability of grippers to adjust their grip to the object being contacted. This work presents a non-camera-based optical tactile sensor array capable of discriminating between textures. The texture discrimination method uses sensor waveform amplitude, morphological, and spectral features. We show that an array of sensing units helps achieve high classification accuracy. Using the phase difference between sensing unit signals as a feature allows for good classification regardless of the scanning speed.