Habitat Simulator

A flexible, high-performance 3D simulator with configurable agents, multiple sensors, and generic 3D dataset handling (with built-in support for MatterPort3D, Gibson, Replica, and other datasets).

When rendering a scene from the Matterport3D dataset, Habitat-Sim achieves several thousand frames per second (FPS) running single-threaded, and reaches over 10,000 FPS multi-process on a single GPU!

The Habitat platform includes Habitat-Sim and Habitat-Lab. To learn how these fit together, see our ECCV 2020 tutorial series.

Tutorials

Basics for Navigation	Video	Jupyter Notebook
Interaction	Video	Jupyter Notebook
Advanced Topics	Video	Jupyter Notebook
Profiling and Optimization	Video	Interactive Colab
New Actions	Page
Attributes Templates JSON Tags	Page
Creating a stereo agent	Page
Working with light setups	Page
View Assets in Habitat-Sim	Page	Jupyter Notebook
Interactive Rigid Objects 2.0	Page	Jupyter Notebook
Gfx Replay		Jupyter Notebook
Editing Scene Assets in Blender	Page
Coordinate Frame Tutorial	Page	Jupyter Notebook

Python Classes

See the Classes tab.

Logging Configuration

See Logging Configuration for how to configure Habitat-Sim logging.

Python Unit Tests

Browse selected unit tests that demonstrate essential Habitat-Sim interfaces.

• test_agent.py
• test_attributes_managers.py
• test_configs.py
• test_controls.py
• test_gfx.py
• test_greedy_follower.py
• test_light_setup.py
• test_navmesh.py
• test_physics.py
• test_pyrobot_noisy_controls.py
• test_semantic_scene.py
• test_sensors.py
• test_simulator.py

C++ API Documentation

Habitat-sim is designed to be used primarily through its Python API. As such, the end-user tutorials and docs linked above focus on Python.

If you’re looking for API reference of Habitat-sim’s C++ internals, please see the C++ API tab.