Environment:Facebookresearch Habitat lab Python 3 9 Core Dependencies

Knowledge Sources	Habitat-Lab habitat-lab setup.py
Domains	Infrastructure, Embodied_AI
Last Updated	2026-02-15 00:00 GMT

Overview

Python 3.9+ environment with NumPy, Hydra, OpenCV, and Gym for running Habitat-Lab core tasks, benchmarking, and configuration composition.

Description

This environment defines the base software stack required to run all Habitat-Lab components. It is built on Python 3.9 or later (3.10, 3.11 also supported) and includes pinned versions of critical numerical libraries (NumPy 1.26.4), Hydra-based configuration management (hydra-core >= 1.2.0, omegaconf >= 2.2.3), and OpenAI Gym (0.22.0 to 0.23.1). The stack supports macOS and Unix platforms. Windows is not officially supported.

Usage

Use this environment for any Habitat-Lab workflow including PointNav training, rearrangement tasks, benchmarking, custom task extension, and HITL interactive evaluation. This is the mandatory base prerequisite for all implementations in the repository.

System Requirements

Category	Requirement	Notes
OS	macOS or Unix/Linux	Windows not officially supported
Python	3.9, 3.10, or 3.11	Listed in setup.py classifiers
Disk	5GB+	For dependencies and dataset caching

Dependencies

System Packages

`python` >= 3.9
`ffmpeg` (for imageio-ffmpeg video generation)
`habitat-sim` (companion simulator, installed separately via conda)

Python Packages

`gym` >= 0.22.0, < 0.23.1
`numpy` == 1.26.4
`numpy-quaternion` >= 2019.3.18.14.33.20
`attrs` >= 19.1.0
`opencv-python` >= 3.3.0
`hydra-core` >= 1.2.0
`omegaconf` >= 2.2.3
`numba` >= 0.44.0
`imageio` >= 2.2.0
`imageio-ffmpeg` >= 0.2.0
`scipy` >= 1.10.1
`tqdm` >= 4.0.0

Credentials

No credentials are required for the core Habitat-Lab environment.

Quick Install

# Install habitat-lab core dependencies
pip install "gym>=0.22.0,<0.23.1" "numpy==1.26.4" "numpy-quaternion>=2019.3.18.14.33.20" \
    "attrs>=19.1.0" "opencv-python>=3.3.0" "hydra-core>=1.2.0" "omegaconf>=2.2.3" \
    "numba>=0.44.0" "imageio>=2.2.0" "imageio-ffmpeg>=0.2.0" "scipy>=1.10.1" "tqdm>=4.0.0"

# Or install via pip from source
cd habitat-lab && pip install -e .

Code Evidence

Python version support from `habitat-lab/setup.py:79-81`:

"Programming Language :: Python :: 3.9",
"Programming Language :: Python :: 3.10",
"Programming Language :: Python :: 3.11",

Platform support from `habitat-lab/setup.py:85-86`:

"Operating System :: MacOS",
"Operating System :: Unix",

Pinned dependencies from `habitat-lab/requirements.txt:1-16`:

gym>=0.22.0,<0.23.1
numpy==1.26.4
numpy-quaternion>=2019.3.18.14.33.20
hydra-core>=1.2.0
omegaconf>=2.2.3
numba>=0.44.0

Pickle5 backward compatibility from `habitat-lab/habitat/utils/pickle5_multiprocessing.py:15-27`:

if sys.version_info[:2] < (3, 8):
    try:
        import pickle5 as pickle
    except ImportError:
        import pickle
        logger.warn(
            f"""Warning pickle v5 protocol not supported.
        Falling back to pickle version {pickle.HIGHEST_PROTOCOL}.
        pip install pickle5 or upgrade to Python 3.8 or greater
        for faster performance"""
        )

Common Errors

Error Message	Cause	Solution
`ModuleNotFoundError: No module named 'habitat_sim'`	habitat-sim not installed	Install habitat-sim via conda: `conda install habitat-sim -c conda-forge -c aihabitat`
`ImportError: numpy.core.multiarray failed to import`	NumPy version mismatch	Pin numpy to 1.26.4: `pip install numpy==1.26.4`
`gym.error.NameNotFound`	Gym version incompatible	Ensure gym >= 0.22.0, < 0.23.1

Compatibility Notes

macOS: Supported but GPU acceleration requires external GPU setup or MPS (limited support).
Windows: Not officially supported. Use WSL2 on Windows.
Python 3.12+: Not listed in classifiers. Known issues with IntFlag behavior in `habitat-hitl` user_mask module.
NumPy: Strictly pinned to 1.26.4. Other versions may cause compatibility issues with habitat-sim.

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment