Visuo Tactile

Visuo-tactile research integrates visual and tactile information to improve robotic manipulation and understanding of object interaction. Current efforts focus on developing robust and efficient models, often employing neural networks (including generative adversarial networks and spiking neural networks) to fuse data from various sensors (e.g., cameras, pressure arrays) and achieve tasks like object pose estimation, shape reconstruction, and contact prediction. This interdisciplinary field is advancing robotic dexterity and perception, with implications for applications ranging from dexterous manipulation in unstructured environments to improved human-computer interaction in virtual reality.