Opinionated. Declarative. Reproducible.

Airfield: The Robotics Engineering & AI Research Framework

Airfield is an opinionated, declarative framework for robotics engineering and AI research. It gives every package an isolated, reproducible runtime environment and brings whole robot stacks up from declarative plans, so a shared system stays reproducible while many people develop packages in parallel.

Quick Install

pipx install git+https://github.com/airfield/airfield.git
airfield system install-completion bash

Why Airfield?

Opinionated Design

Stop wrestling with boilerplate code and configurations. Airfield makes code and dependency management easy, so you can focus on building the algorithms that enable perception, planning, and control.

Reproducibility and Encapsulation

Airfield locks and isolates your package dependencies to allow your code to run on your laptop, in CI, and on the physical robot or training cluster, even when multiple developers are working on the same system.

Declarative Orchestration

Bring up your robot stack from declarative package plans. Airfield keeps package boundaries explicit while generating repeatable launch workflows for development and productionization.

Explore Airfield

Understand the Model

How Airfield wraps ROS packages into containers: the full architecture in six diagrams, from the command surface to launch time.

Start Building

Initialize a project, build a reproducible image, and launch a robot package, then follow the tutorials for each piece in depth.

CLI Reference

Every command and flag, the airfield.yaml, plan, and dependency manifest schemas, plus how containers are wired at runtime.