Active Tactile Exploration for Rigid Body Pose and Shape Estimation
Ethan Kroll Gordon, Bruke Baraki, Hien Bui, Michael Posa
AI summary
Problem
Tactile sensing suffers from data sparsity and object disturbance during contact, making it difficult to efficiently learn object geometry and pose without prior models or visual data.
Approach
The method simultaneously estimates pose and shape using only tactile data by optimizing a violation-implicit loss for physical constraints and selecting actions to maximize Expected Information Gain.
Key results
- Simultaneous pose and geometry learning from purely tactile data
- Violation-implicit loss enables stable gradient-based optimization
- EIG-driven exploration accelerates learning in simulation and real robots
- Accurate estimation achieved with under 10 seconds of post-contact data
Why it matters
Enables robots to efficiently build accurate physical models of unseen, dynamic objects using only touch, improving manipulation robustness in occluded or cluttered environments.
Abstract
General robot manipulation requires the handling of previously unseen objects. Learning a physically accurate model at test time can provide significant benefits in data effi- ciency, predictability, and reuse between tasks. Tactile sensing can compliment vision with its robustness to occlusion, but its temporal sparsity necessitates careful online exploration to maintain data efficiency. Direct contact can also cause an unrestrained object to move, requiring both shape and location estimation. In this work, we propose a learning and exploration framework that uses only tactile data to simultaneously deter- mine the shape and location of rigid objects with minimal robot motion. We build on recent advances in contact-rich system identification to formulate a loss function that penalizes physical constraint violation without introducing the numerical stiffness inherent in rigid-body contact. Optimizing this loss, we can learn cuboid and convex polyhedral geometries with less than 10s of randomly collected data after first contact. Our explo- ration scheme seeks to maximize Expected Information Gain and results in significantly faster learning in both simulated and real-robot experiments. More information can be found at: https://dairlab.github.io/activetactile.