Sign in to view source links and access this dataset
Description
NVIDIA's PhysicalAI DigitalCousin Assets provide a collection of 3D meshes, textures, and object metadata for simulated tabletop manipulation environments. These digital assets, including mugs, bottles, bowls, and containers, populate virtual scenes for the GR1 robot. The dataset was published by NVIDIA and last updated in June 2025.
Use Cases
Training robotic manipulation policies using 3D mesh geometry and object metadata for physics simulation.
Benchmarking robot performance on tabletop tasks with standardized assets like mugs and bottles.
Developing computer vision models for object recognition and pose estimation from provided textures and 3D models.
Creating synthetic datasets for sim-to-real transfer by rendering scenes with these asset collections.
Studying grasp planning and object interaction using the physical properties inferred from the asset metadata.
Strengths
Assets are designed for a specific, high-fidelity robotics simulation platform (GR1).
Collection includes multiple asset types (mugs, bottles, bowls, containers) for diverse scene composition.
Limitations
Unknown total number of assets, meshes, or textures limits assessment of scope.
Lack of column or metadata schema details prevents understanding of specific object properties.
Potential bias towards tabletop objects may limit use for other robotic manipulation domains.
Provenance
Source
NVIDIA
Collection Method
Digitally created assets for simulation.
Freshness
Updated June 2025.
Full description and access details are on the external Hugging Face dataset page; license and specific file formats are unknown from provided input.