NUST Institutions Library Catalogue

Vision-based tactile sensors provide an effective solution for robotic manipulation tasks
requiring high sensitivity, such as grasping fragile objects. This research presents a tactile
sensing approach based on a flat elastic skin embedded with visual markers and integrated with
a depth camera on a parallel robotic gripper. A set of algorithms has been developed that process
marker displacements to estimate distributed contact forces and contact areas during object
interaction. These estimations are then utilized in computing distributed tactile pressure values
and for detecting incipient slip through pressure variation analysis. The main contribution of
this research is the development of a unified framework that integrates distributed force
estimation with force-based contact detection and pressure-based slip detection for improved
grasping of fragile objects. Specifically: (1) a marker tracking algorithm is proposed to detect
surface deformation from the tactile skin, enabling distributed force estimation; (2) the
estimated forces are used to compute pressure distribution maps and identify contact regions;
and (3) a slip detection algorithm is introduced that leverages changes in the pressure field to
reliably detect incipient slip during manipulation. Experimental results demonstrate that the
proposed method achieves accurate slip detection with 98.9% success rate, and improves grasp
reliability, achieving 99.5% success in real-time manipulation tasks. The integration of
distributed force, pressure, and slip estimation significantly enhances robotic grasp stability
and enables safe handling of delicate items.