GelStereo BioTip: Self-Calibrating Bionic Fingertip Visuotactile Sensor for Robotic Manipulation

摘要

High-resolution visuotactile sensors are often bulky and have flat contact shapes, making it difficult to integrate into dexterous multifinger fingertips and perform manipulation tasks smoothly. In this article, we present a soft, low-cost, biomimetic fingertip-sized visuotactile sensor, which can perceive high-precision 3-D surface deformation. This novel sensor, named GelStereo BioTip, adopts stereo vision system to robustly acquire 3-D deformation information on multicurvature contact surfaces. In particular, a self-calibrating refractive stereo ray tracing modeling method is proposed to solve the problem of ray refraction on multimedium and multicurvature refractive surfaces in the binocular reconstruction system. GelStereo BioTip has an overall spatial resolution of 1.2 mm, a shape measurement accuracy of less than 0.28 mm over a range of 0–5 mm for numerous distinct contacts with varying contact area. We integrate the bionic fingertips into a dexterous robotic hand for validation, and the proposed bionic fingertip concept can be seamlessly transferred to a variety of robotic manipulators.