R-Tac: A Rounded Monochrome Vision-based Tactile Sensor

Wanlin Li, Pei Lin, Meng Wang, Chenxi Xiao, Kaspar Althoefer, Yao Su, Ziyuan Jiao, Hangxin Liu

Keywords: Tactile sensor design

TL;DR: In this work, we introduce R-Tac, a low-cost rounded monochrome vision-based tactile sensor engineered for high-resolution and high-speed perception.

Abstract:

Abstract— Endowing the curved surfaces of rounded vision-
based tactile sensors (VBTS) is essential for dexterous robotic
manipulation, as they offer more sufficient contact with the
environment. However, current rounded designs are constrained
by a low sensing frequency (30–60 Hz) and the need for
extensive calibration when adapting to new sensors due to
the reliance on multi-channel captures, which hinders their
performance in dynamic robotic tasks and large-scale deployment. In this work, we introduce R-Tac, a low-cost
rounded VBTS engineered for high-resolution and high-speed
perception. The key innovation is a monochrome vision-based
sensing principle: utilizing a single-channel camera to capture
the reflection properties of the compound rounded elastomer
under monochromatic illumination. This single-channel sensing
principle significantly reduces data volume and simplifies computational complexity, enabling 120 Hz tactile perception. R-Tac
features an efficient calibration process, requiring only a few
captures with a 3D-printed calibration setup. We demonstrate
the advantages of R-Tac’s design through a series of tasks including object grasping, paper picking, in-
hand object reorientation, and terrain exploration, showcasing
its effectiveness for dexterous robotic hands.

Submission Number: 14