1,000 3D object models featuring synchronized visual, acoustic, and tactile data. The collection includes 3D meshes, simulated impact sounds, and high-resolution tactile images generated via the DIGIT sensor simulation.
Use Cases
- Train cross-modal models to predict 3D mesh geometry from acoustic impact sounds
- Develop robotic grasping algorithms that utilize tactile images and contact force vectors for feedback
- Perform material classification using the provided density and Young's modulus labels
Strengths
- 1,000 3D objects with high-fidelity meshes and textures derived from ShapeNet
- Acoustic data containing impact sounds and impulse responses for 100 contact points per object
- Tactile data providing high-resolution touch images and contact force maps
- Material parameters including density, Young's modulus, and Poisson's ratio for physics-based modeling