| --- |
| license: cc-by-4.0 |
| language: |
| - en |
| tags: |
| - robotics |
| - reinforcement learning |
| - embodied ai |
| - computer vision |
| - simulation |
| size_categories: |
| - 1M<n<10M |
| task_categories: |
| - reinforcement-learning |
| - robotics |
| viewer: false |
| --- |
| |
| # AI2THOR Scene/Object Dataset for ManiSkill |
|
|
| This is a modified version of the [Habitat Synthetic Scenes Dataset (HSSD)](https://huggingface.co/datasets/hssd/ai2thor-hab) which is also modified based on the original [AI2THOR](https://github.com/allenai/ai2thor/) scenes and assets |
| in order to make it usable by [ManiSkill](https://github.com/haosulab/ManiSkill). There are 10,000+ scenes and 10,000+ different objects so downloading this can take a bit of time. |
|
|
| Note that currently this dataset has no interactable articulated objects as the original AI2THOR simulator is based on unity and had magic articulation scripts that were not / could not be ported into usable urdfs. |
|
|
| Changes to HSSD for repo: |
| - Add decomposed collision meshes for a number of objects, although not all are decomposed yet. |
|
|
| After installing ManiSkill to download this dataset run |
| ``` |
| python -m mani_skill.utils.download_asset "AI2THOR" |
| ``` |
