AI Habitat Resource Index

Context

Understand project context and organization.

About AI Habitat [homepage]

Platform Architecture: See how the pieces fit together.

• sim: https://github.com/facebookresearch/habitat-sim/blob/main/DETAILS.md
• lab: https://github.com/facebookresearch/habitat-lab/blob/main/DETAILS.md
• spot sim-2-real: https://github.com/facebookresearch/spot-sim2real

Concepts and Terminology [page]: Browse a dictionary of project terms and concepts.

Paper Links (expand): Read and cite Habitat conference papers.

• Habitat 3.0: A Co-Habitat for Humans, Avatars and Robots [arxiv] (2023)
• Habitat 2.0: Training Home Assistants to Rearrange their Habitat [arxiv] (2021)
• Habitat: A Platform for Embodied AI Research [arxiv] (2019)

Community

Reach the developers, ask questions, discuss ideas, and contribute.

Discussion Forum [link]: ask questions, share your ideas, connect with the community.

Reporting Issues (habitat-lab | habitat-sim): search for answers, report a bug, get support, request a feature.

Contributing to AI Habitat (expand): we welcome your contributions!

• Before we can accept your code contributions, we need you to submit a Contributor License Agreement (CLA). You only need to do this once to work on any of Meta’s open source projects.
• 🖊️ Complete your CLA here.

Steps to contributing:

• First, check out the contributing guidelines for habitat-lab and the developer workflow for habitat-sim to brush up on processes, style guides, and developer tips.
• Then fork the repo you will be modifying and make any changes you want!
• When you are ready, submit your code for review by opening a pull request (PR). See the links above for details on PR submission in each repo. Prepare to accept developer feedback and iterate before merging.

Documentation

Browse documentation resources on a variety of subjects.

Habitat-Lab [doc homepage] [Python API]

Habitat-Sim [doc homepage] [Python API] [C++ API]

Dataset Docs (expand): fuel your experiments.

• Habitat-sim Supported Asset Datasets [readme]
• Habitat-lab Scene and Episode Datasets [readme]
• Habitat SceneDatasets [page]: configure 3D assets and metadata via JSON and programmatic APIs.

Miscellaneous Topics (expand)

• Habitat-sim Configurable Logging [page]: get only the logs you want to see from Habitat-sim.
• Habitat-lab Configurable Logging [page]: TODO
• Editing 3D scene Assets in Blender [page]: learn about editing 3D scene assets in Blender for visualization.

Tutorials

Learn hands-on with runnable python scripts and interactive Jupyter notebook tutorials.

Habitat-lab Tutorials (expand)

• Quickstart [page]: Jump into the basics with the pointnav task.
• Demo [page]: Configs, scene semantics, actions, and sensors.
• Top Down Map Visualization [page]: Visualizing a top down map of the scene.

Habitat-sim Tutorials (expand)

• Rigid Object API [page]: Learn to instance and control rigid objects.
• Articulated Object API [page TODO] : Learn to instance and control articulated objects (hinged and jointed objects).
• Customizing Light Setups [page]: Learn to customize scene lighting.
• Asset Viewer [page]: View 3D assets interactively in Habitat-sim.
• Basics for Navigation [Jupyter notebook]: Learn the Habitat-sim basics for embodied navigation: simulator initialization, sensors, agents, navigation mesh, topdown map, taking actions.
• Basics for Interaction [Jupyter notebook]: Learn how to use Habitat-sim for interactive simulations: rigid object API, physics simulation, sampling object locations, continuous control.
• Advanced Topics [Jupyter notebook]: Learn about more advanced Habitat-sim features such as: tracking object motion with a camera, projecting and unprojecting points, object semantic ids, programmatically manipulating object metadata.
• GFX Replay [Jupyter notebook]: Learn how to record and replay simulations.

Miscellaneous Topics (expand)

• Profiling and Optimization [video] [Colab non-running reference]: Learn strategies for profiling and optimizing your experiment code.
• Understanding Habitat’s Coordinate System [page]
• Create a Stereo Agent [page]
• Image Extractor [page]: Using the ImageExtractor tool to produce offline image datasets from synthetic scenes.
• Habitat 2.0 Overview [page]: Learn about habitat 2.0 tasks and configuration.

ECCV 2020 Tutorial Series [page]: Archived ECCV 2020 tutorial series. Original Colabs and videos available. Updated tutorial notebooks linked above.

More from AI at Meta

Explore more exciting projects from the Fundamental AI Research (FAIR) team and AI at Meta.

https://ai.meta.com/research/