For robots doing real-world manipulation tasks it is very useful to measure the force in the finger tips to determine grasping force at a fine grained level, and also the shape of the object being grasped. Tactile sensing has a long history in Robotics, and recent advances in tactile sensors and machine learning look promising to help better perceive and understand contact forces. This project will mainly explore different tactile sensors and algorithms/methods to understand their signals. Some exemplary research questions include:
- how tactile information can help object property estimation and robotic manipulation,
- how various interaction motions will influence the sensing results,
- and how interaction and tactile sensing can be better coordinated in an active manner for higher performing object property estimation.