InternScenes: Large-Scale Interactive Indoor Scenes for Embodied AI

InternRobotics released InternScenes in 2025, a large-scale collection of interactive indoor scenes featuring realistic layouts for embodied AI research. Presented at NeurIPS 2025, the data focuses on scene generation and robotic interaction within complex 3D environments.

Use Cases

• Training embodied AI agents to perform object interaction using realistic layout features
• Developing scene generation models that prioritize functional spatial arrangements
• Evaluating robotic navigation algorithms in multi-object indoor environments

Strengths

• NeurIPS 2025 peer-reviewed publication status
• Focus on interactive elements rather than static geometry
• Realistic indoor layout configurations for embodied AI

Limitations

• Lack of public documentation on specific scene counts
• Geographic or architectural bias toward standard indoor layouts
• Unknown file format compatibility for non-Robotics simulators

Provenance

Source

InternRobotics

Freshness

Current as of its NeurIPS 2025 publication date.