Utilization of Skin Color Change for Image-based Tactile Sensing
🤖 AI Summary
Existing vision-based tactile methods rely on deformation sensing, making them vulnerable to occlusion and illumination variations, and incapable of directly capturing fingertip pressure distribution. This work proposes a novel image-based tactile paradigm grounded in skin optical chromaticity shift—rather than mechanical deformation—induced by contact forces. Specifically, high-resolution cameras capture minute CIELAB color shifts in a soft, elastomeric tactile skin under load; a physics-informed chromaticity–pressure mapping model is then established. By integrating multispectral imaging, a lightweight neural network, and calibration-compensation algorithms, the approach achieves high-accuracy, sensor-free force estimation. It attains a prediction error of ±0.08 N across a 0–5 N range, a spatial resolution of 2 mm, and exhibits threefold greater robustness to illumination interference compared to state-of-the-art vision-based tactile methods. This framework establishes a high-fidelity, robust pathway for tactile perception in robotic teleoperation and natural human–robot interaction.
Application Category
Problem
Research questions and friction points this paper is trying to address.
Measure fingertip pressure distribution via skin color changes
Address limitations of indirect pressure sensing methods
Analyze spatial non-uniformity in color-pressure correlation
Innovation
Methods, ideas, or system contributions that make the work stand out.
Measures pressure via skin color change
Uses transparent object for direct touch
Analyzes stress with finite element method
🔎 Similar Papers
No similar papers found.
