Vision-Based Tactile Sensing
Overview
Vision-based tactile sensors image soft surfaces as they deform, letting robots estimate contact shape, force, slip, texture, and material properties. Methods include marker tracking and photometric reconstruction, learned models for tactile images and time sequences, and physics-based simulation or synthetic data generation. These signals support tasks such as grasping, insertion, cloth handling, palpation, and contact-rich control, with growing work on compact sensors and low-latency processing. A central challenge is reliable transfer across sensor designs and real-world conditions: calibration, sensor variation, wear, and limited labeled data can still reduce performance.
Latest papers
60 papers8 papers in the last four weeks, level with the four weeks before. 0.0% of all new papers.