Deformation Pattern Shape Sensing for Visuotactile Endoscope

Yanzhe Yu,Yuan Tian,Shaowei Cui,Yupeng Wang,Rui Wang,Shuo Wang

DOI: https://doi.org/10.1109/rcar61438.2024.10671278

2024-01-01

Abstract:The integration of visuotactile sensor technology with endoscopy enables the provision of tactile information for minimally invasive surgeries and other operations. This paper proposes a real-time sensing framework for contact deformation pattern shape, providing a foundation for precise perception in miniaturized visuotactile sensors. Specifically, we first construct a Deformation Pattern shape Descriptor (DPSD) and then track the key points in the descriptor using optical flow to obtain contact deformation shape information. Through extensive qualitative and quantitative experiments, our proposed algorithm achieves an average perception accuracy of around 83.81% for contact pattern shapes. It demonstrates consistent perception accuracy across various contact force magnitudes, directions, and scenarios, highlighting its robustness. Additionally, compared to traditional marker-level methods, it achieves a notable 46.5% improvement in displacement field perception accuracy. This research elevates tactile perception to the dimension of contact pattern shape sensing and has the potential for application in miniaturized visual tactile sensors.