A Mesh-Type Topology Pattern for Robust Marker Tracking in Visuotactile Sensor

Visuotactile sensors (VTS) perceive the tactile information through the elastomer deformation captured by the embedded camera. Markers are commonly added to the elastomer surface for VTS to get physical process sensing capabilities, whatever the imaging environment. Consequently, the success of marker tracking is a prerequisite for VTS perception. Marker missing and mismatching become one of the biggest limitations of VTS performance improvement. In this paper, a mesh-type topology pattern-based frame-independent marker tracking method is proposed to enhance the tracking robustness, particularly when large deformation occurs. The mesh pattern provides connectivity between different markers and the corresponding algorithm fully utilizes the inherent topological relations of all neighboring markers within the mesh for marker tracking. The effectiveness of the proposed method is validated both numerically and experimentally. The tracking accuracy is qualitatively analyzed by comparing it with traditional tracking methods and further evaluated based on a VTS prototype. Results confirmed that the proposed method significantly enhances the tracking robustness under large deformations and its compatibility with texture perception in photometric stereo environment.