Active Visuo Tactile

Active visuo-tactile perception research focuses on enabling robots to efficiently learn and perform manipulation tasks by integrating visual and tactile sensing. Current efforts concentrate on developing algorithms, such as reinforcement learning coupled with active inference and Bayesian filtering, to guide robots in actively exploring objects, estimating their properties (e.g., mass, shape), and executing precise actions like grasping and insertion. This research is significant because it addresses limitations of purely vision-based approaches in cluttered or uncertain environments, paving the way for more robust and adaptable robotic manipulation in real-world applications. The development of efficient and generalizable models for active visuo-tactile perception is crucial for advancing robotics in areas such as manufacturing, logistics, and assistive technologies.