244K embodied robotics simulation demonstrations contain frame-based information including 2D/3D boxes, trajectories, grasp points, and semantic masks. The dataset was created by InternRobotics and last updated on December 12, 2025.
Use Cases
- Training robotic manipulation models based on annotated grasp points.
- Developing trajectory prediction algorithms based on trajectory data.
- Evaluating object detection and segmentation models based on 2D/3D boxes and semantic masks.
- Simulating robotic tasks based on comprehensive frame-based demonstrations.
Strengths
- 244K demonstrations provide a substantial volume of training examples.
- Annotations include multiple modalities such as 2D/3D boxes, trajectories, grasp points, and semantic masks.
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown, which may limit suitability assessment.
- Description metadata is limited; actual data quality requires manual inspection after download.
Provenance
- Source
- InternRobotics
- Collection Method
- Simulation demonstrations.
- Freshness
- Last updated 2025-12-12 17:44:49; freshness should be verified.