JacopoPan/aerial-autonomy-stack

What is it?

Key features

• 01Simulates PX4, ArduPilot and Betaflight multi-vehicle setups with quadrotors, VTOLs and tailsitters.
• 02Provides ROS2 action interfaces for takeoff, land, orbit and offboard control.
• 03Runs YOLO detection and LiDAR odometry with ONNX GPU runtimes.
• 04Offers a Gymnasium environment and faster-than-real-time, multi-instance simulation.
• 05Supports Dockerized simulation and deployment on NVIDIA Jetson Orin.
• 06Includes Zenoh-based inter-vehicle communication and a dual network for simulated sensors and vehicle comms.

Use cases

• Prototyping multi-drone formation or swarm behaviors before flight tests.
• Training or evaluating reinforcement learning policies with a Gymnasium interface.
• Testing perception pipelines such as detection and LiDAR odometry in simulated 3D worlds.
• Deploying onboard perception and control code to Jetson-based drones.

Quick start

1. 1

   Install git, xterm, xfonts-base, wget and unzip

   sudo apt update && sudo apt install -y git xterm xfonts-base wget unzip
2. 2

   Clone the repository and enter the tools folder

   git clone https://github.com/JacopoPan/aerial-autonomy-stack.git && cd aerial-autonomy-stack/tools_and_docs/
3. 3

   Check that NVIDIA driver, Docker and nvidia-container-toolkit are installed

   ./tests/check_requirements.sh
4. 4

   Build the simulation images (first build takes about 45 minutes)

   ./sim_build.sh

Things to know

• !Requires an NVIDIA GPU with driver, Docker and nvidia-container-toolkit.
• !Developed and tested on Ubuntu 24.04; Windows 11 needs WSL.
• !The first build takes roughly 45 minutes.

In collections

Topics

FAQ

What is aerial-autonomy-stack?

An open ROS2 framework to simulate and deploy perception-based drone swarms across PX4, ArduPilot and Betaflight autopilots. Aerial autonomy stack (AAS) is a containerized software harness for developing multi-drone autonomy. It runs PX4, ArduPilot and Betaflight flight controllers in simulation with Gazebo, covering quadrotors, quadplane VTOLs and tailsitters, and exposes a ROS2 action-based interface to control them.

Is aerial-autonomy-stack free to use? What license does it have?

aerial-autonomy-stack is open source under the MIT license. Check the license terms before using it in commercial products.

Is aerial-autonomy-stack still maintained?

Actively maintained. The last commit to aerial-autonomy-stack was on Nov 6, 2026.

What can I use aerial-autonomy-stack for?

Prototyping multi-drone formation or swarm behaviors before flight tests. Training or evaluating reinforcement learning policies with a Gymnasium interface. Testing perception pipelines such as detection and LiDAR odometry in simulated 3D worlds. Deploying onboard perception and control code to Jetson-based drones.

How do I install aerial-autonomy-stack?

Install git, xterm, xfonts-base, wget and unzip: sudo apt update && sudo apt install -y git xterm xfonts-base wget unzip Clone the repository and enter the tools folder: git clone https://github.com/JacopoPan/aerial-autonomy-stack.git && cd aerial-autonomy-stack/tools_and_docs/ Check that NVIDIA driver, Docker and nvidia-container-toolkit are installed: ./tests/check_requirements.sh Build the simulation images (first build takes about 45 minutes): ./sim_build.sh

Where can I get aerial-autonomy-stack?

The source code and installation instructions are on GitHub: https://github.com/JacopoPan/aerial-autonomy-stack