Set up isolated Python environments for OpenClaw development using venv, conda, and Poetry. Avoid dependency conflicts, manage multiple robot projects cleanly, and reproduce environments identically across team machines and CI servers.
Ensure Python 3.10 or later is installed (python3 --version). Create a virtual environment: python3 -m venv openclaw-env. Activate it: source openclaw-env/bin/activate (Linux/macOS) or openclaw-env\Scripts\activate (Windows). Install OpenClaw: pip install openclaw. Verify: python -c "import openclaw; print(openclaw.__version__)". Deactivate when done: deactivate. Store the environment outside your project directory to avoid accidentally committing it to version control.
Create a new conda environment with the correct Python version: conda create -n openclaw-dev python=3.11 -y. Activate: conda activate openclaw-dev. Install OpenClaw from PyPI (not conda-forge, as the openclaw package is only on PyPI): pip install openclaw. Install conda-managed scientific dependencies: conda install numpy scipy -c conda-forge. Verify: conda list | grep openclaw. Save the environment: conda env export > environment.yml for reproducibility.
Install Poetry: curl -sSL https://install.python-poetry.org | python3 -. Create a new project: poetry new my-robot-project && cd my-robot-project. Add OpenClaw: poetry add openclaw. Add optional extras: poetry add "openclaw[vision]" or poetry add "openclaw[ai]". Run scripts: poetry run python my_script.py. Lock the environment: poetry lock. Commit both pyproject.toml and poetry.lock to version control so all collaborators use identical dependency versions.
OpenClaw requires Python 3.10 or later. Python 3.11 is the recommended version as of OpenClaw 3.1 for best performance and broadest library compatibility. Python 3.12 is supported but some third-party robotics libraries have not yet released compatible wheels. Python 3.9 and earlier are not supported. Run python3 --version or python --version to verify your installed version before creating a virtual environment.
After activating your environment (venv, conda, or Poetry shell), run: pip install "openclaw[vision]". This installs openclaw plus opencv-python, numpy, and Pillow. For full AI integration: pip install "openclaw[ai]" which adds PyTorch and torchvision. For all extras: pip install "openclaw[all]". Note: PyTorch installation via pip installs the CUDA-enabled wheel on Linux by default; specify pip install "openclaw[ai]" --index-url https://download.pytorch.org/whl/cpu for CPU-only.
With venv: generate a requirements file: pip freeze > requirements.txt. Team members recreate with: python3 -m venv openclaw-env && source openclaw-env/bin/activate && pip install -r requirements.txt. With conda: conda env export > environment.yml; team recreates with: conda env create -f environment.yml. With Poetry: commit pyproject.toml and poetry.lock; team installs with: poetry install. Poetry is the most reliable for exact reproducibility; conda is best when binary (non-pure-Python) dependencies are involved.
Create a fresh isolated environment for the conflicting project: python3 -m venv fresh-env && source fresh-env/bin/activate. Install only the packages needed for that project. Run openclaw doctor --python inside each environment to identify conflicts. Common conflicts involve numpy version ranges: openclaw requires numpy>=1.24 while some older ROS Python tools pin numpy==1.21. The solution is always to use separate virtual environments rather than trying to resolve conflicting pins in a shared environment.
Yes. Install Jupyter in your activated environment: pip install jupyterlab. Register the environment as a Jupyter kernel: pip install ipykernel && python -m ipykernel install --user --name openclaw-env --display-name "OpenClaw (Python 3.11)". Launch Jupyter: jupyter lab. Select the "OpenClaw (Python 3.11)" kernel in the notebook. This lets you interactively control a robot or simulator from a notebook cell without switching environments.
Integrate OpenAI GPT-4o, Anthropic Claude 3.5 Sonnet, Google Gemini 2.0 Flash, and self-hosted Ollama LLaMA 3 with OpenClaw for natural-language robot instruction and AI-powered manipulation planning. Includes detailed per-call and monthly cost breakdowns for each AI provider.
Deploy OpenClaw on AWS EC2 and EKS, Azure Virtual Machines and AKS, and Google Cloud Compute Engine and GKE. Includes detailed cost breakdowns per cloud, GPU instance sizing for simulation, and architecture patterns for hybrid cloud-to-robot connectivity.
Deploy OpenClaw inside Docker containers for reproducible, portable robotics development. This guide covers single-container setups, multi-service docker-compose stacks, GPU passthrough for simulation, and production image hardening.