---
license: mit
---

# MaGalRose Series

```
The MaGalRose series is a collection of interactive robot manipulation datasets and simulation
environments designed to model real-world manipulation behaviors with higher fidelity and greater
environmental diversity. Building upon the real-world manipulation trajectories and multimodal
observations of MaNI, MaGalRose expands the scale and complexity of interactive environments,
introducing a large number of manipulation scenarios with diverse structures, semantic variations,
and finely modeled physical properties. This allows robots not only to reproduce actions from
existing examples but also to actively explore, experiment, and adjust strategies within the
environment. Through more refined geometric modeling, physical parameter calibration, and
contact process modeling, the dataset significantly approximates the real world in terms
of visual appearance, dynamic response, and interaction consistency, thus enabling more
natural, stable, and generalizable manipulation behavior learning. Designed with the
beauty and completeness symbolized by "Rose" in mind, MaGalRose combines high-quality
modeling with excellent interactivity, providing an expressive and usable experimental
platform for research into long-term manipulation planning, multi-environment
generalization, and sim-to-real transfer.
```

MODAL : Data Confuser "Small files only, Big files confusion"

CHARS : 8921253f79544x260235411p

DCVERSION : 0.0.1.2 built 260210