206 digital-twin articulated objects integrated into 12 distinct scenes, including 6 pre-configured and 6 user-defined environments. The data features pixel-level affordance annotations and modular interaction components designed for physics-based simulations.
Use Cases
- Train computer vision models to predict interaction zones using the pixel-level affordance annotations
- Simulate robotic grasping and manipulation tasks using the 206 articulated digital-twin objects
- Benchmark spatial reasoning in 3D environments using the 12 pre-configured and user-defined scenes
Strengths
- 206 high-quality digital-twin articulated objects
- 12 digital-twin scenes consisting of 6 pre-configured and 6 user-defined setups
- Pixel-level affordance annotations for precise interaction mapping
- Modular interaction framework built for physics fidelity