Project AirSim

Project AirSim

Latest release Ubuntu Unreal Engine ROS 2 Python License: MIT IAMAI Simulations Discord

Project AirSim is an open-source, extensible, engine-independent simulation platform for autonomous systems. Its simulation core and APIs can run in the lightweight Project AirSim Runtime without Unreal Engine, or with Unreal Engine 5 when a 3D world, rendered sensors, and environment geometry are required.

Microsoft AirSim → Microsoft Project AirSim → IAMAI Project AirSim

Microsoft developed Project AirSim as the evolution of AirSim. After Microsoft discontinued the Project AirSim initiative, engineers who had worked on Microsoft AirSim and Project AirSim took the initiative to carry the project forward at IAMAI Consulting Corp.. Today, IAMAI maintains and extends this open-source project, bringing that engineering experience to its continued development alongside the community.

Read the Project AirSim announcement in Microsoft’s AirSim repository for the project’s origins and Microsoft’s acknowledgment of IAMAI’s contributions.

Project AirSim provides a modular framework for drones, fixed-wing aircraft, unmanned ground vehicles (UGVs), and other autonomous systems.

Download the latest release · Use a pre-built environment · Build from source · Read the documentation

Autonomous landing application with a live camera view
Autonomous Landing. Perception-guided vehicle control with a live camera stream.		 Quadrotor flying through adjustable weather conditions
Adjustable Weather. Change environmental conditions while the simulation is running.
Fixed-wing swarm flying over a mountain environment
Simulate Your Swarm. Run multiple vehicles together in a shared simulation.		 Air taxi flying over a dynamic city
Dynamic City. An air taxi operating in a dense Unreal city environment.
Aircraft inspecting wind turbines
Wind Turbine Inspection. Inspect renewable-energy infrastructure in a large Unreal environment.		 Caravan of unmanned ground and aerial vehicles traversing a mountain village
Multi-Vehicle Caravan. Simulate UGVs and aerial vehicles together with multiple sensor integration.

Table of Contents

Current Repository Capabilities

The current main branch lets you:

  • Create configurable single-vehicle and multi-vehicle simulation scenes.

  • Use built-in C++ Fast Physics, JSBSim flight dynamics, Simulink physics models, or extend the C++ physics layer for custom requirements.

  • Use Simple Flight, PX4 in software-in-the-loop (SITL) or hardware-in-the-loop (HITL), ArduPilot SITL, and manual control workflows.

  • Integrate custom controllers, actuators, sensors, and robot models.

  • Simulate fixed-wing aircraft with JSBSim, including Cessna 310 and Skywalker X8 examples.

  • Simulate UGVs through Unreal Blueprint vehicle integration, including the SimpleDrive SUV example.

  • Connect autonomy software through Python and C++ client libraries or the ROS 2 C++ bridge.

  • Work with cameras, LiDAR, radar, IMU, GPS, barometer, magnetometer, airspeed, and other configurable sensors.

  • Run with Unreal rendering, off-screen rendering, or without rendering for controller, API, and physics development.

Published binaries may trail the main branch. Check the release notes to confirm which capabilities are included in a particular package, and see the changelog for repository changes.

Choose Your Starting Point

Run a Pre-built Environment

I want to evaluate Project AirSim, launch an environment, and control a vehicle with Python.

Download a packaged environment from GitHub Releases, then follow the pre-built environment guide.

Build and Extend Project AirSim

I want to customize the simulation core, Unreal plugin, vehicles, sensors, or environments.

Follow the source development guide to build the simulation libraries, plugin, Blocks environment, and client packages.

Run Without Unreal Engine

I want to run a simulation without Unreal Engine or rendered sensors.

Use Project AirSim Runtime, the lightweight, engine-independent host. It runs the same SimServer, Project AirSim physics, controllers, APIs, and non-rendered sensors without requiring Unreal Engine.

Runtime supports Fast Physics, JSBSim, Simulink physics, flight-controller workflows, and sensors including GPS, IMU, barometer, magnetometer, and airspeed. It does not provide cameras, LiDAR, radar, Unreal world meshes, or general mesh and robot-to-robot collisions. Its host-side collision support is limited to a flat ground plane for Fast Physics vehicles.

Migrate from AirSim

I have an existing AirSim environment or client workflow.

Start with the AirSim transition guide.

Latest Project Updates

The current main branch contains the changes recorded for Project AirSim 1.0.0, including:

  • JSBSim fixed-wing simulation with Cessna 310 and Skywalker X8 examples;

  • a standalone C++ client package and ROS 2 C++ bridge;

  • GPU LiDAR 360-degree scanning and additional LiDAR validation;

  • a Python client step() API; and

  • Project AirSim Runtime and modular scene configurations;

  • Blueprint vehicle integration, including SimpleDrive SUV examples; and

  • improved build, toolchain, and CI support for Unreal Engine 5.7.

See Project AirSim releases for the latest published binaries and release notes.

Architecture

Project AirSim has three primary layers:

  1. Simulation libraries provide the base infrastructure for defining robot structures, physics, controllers, sensors, and the simulation scene tick loop.

  2. Simulation host provides the environment-dependent services. Project AirSim Runtime offers lightweight engine-independent execution, while the Unreal Engine plugin adds 3D environments, rendering, mesh interaction, and rendered sensors.

  3. Client libraries expose network APIs for loading scenes, controlling vehicles, and receiving state and sensor data.

Project AirSim architecture: clients connect to either the lightweight Runtime host or the Unreal host, both of which use the common simulation libraries

For more detail, see the Project AirSim architecture overview.

Experimental Unity Host Reference

The repository also contains an experimental Unity host integration composed of a Unity example project and a native wrapper around the Project AirSim simulation libraries. It is retained as reference code that demonstrates how another 3D engine can host the common simulation core.

The Unity integration is not currently maintained, validated, packaged, or supported by IAMAI and is not part of the supported-platform matrix. Do not assume compatibility with current Project AirSim or Unity releases.

Key Integrations and Reference Documentation

Supported Development Platforms

The supported development baseline is:

Component

Supported version or behavior

Linux

Ubuntu 22.04 is the primary supported distribution

Windows

Windows 11 with Visual Studio 2022 C++ build tools

macOS

Apple Silicon (ARM64), macOS 15 CI: SimLibs Release build/tests and headless Runtime build; see Mac setup and scope

Unreal Engine

5.2, 5.7, or 5.8

CMake and C++

CMake 3.15 or newer and C++17

Linux compiler

Unreal’s packaged toolchain when UE_ROOT is set; otherwise Clang 13

Windows compiler

build.cmd selects MSVC 14.37 for UE 5.2 and MSVC 14.44 for UE 5.7 and 5.8

Python client

Python newer than 3.10, below Python 4

ROS 2 C++ bridge

ROS 2 Humble on Ubuntu 22.04

setup_linux_dev_tools.sh recognizes some additional Ubuntu releases, but that installation logic is not a supported-platform guarantee. Use Ubuntu 22.04 for the documented and CI-tested Linux development environment.

Hardware requirements are primarily determined by Unreal Engine and the rendering workload. Review the system specifications before installing or building the project.

Source Build Overview

The complete and authoritative instructions are in the source development guide. The basic workflow is summarized below.

1. Install Unreal Engine

Install Unreal Engine 5.2, 5.7, or 5.8 and set UE_ROOT to its installation path.

On Linux:

export UE_ROOT=/path/to/UnrealEngine

2. Install Linux Development Dependencies

./setup_linux_dev_tools.sh

3. Build the Simulation Libraries

On Linux:

./build.sh simlibs_debug

On Windows, use an x64 Native Tools Command Prompt for VS 2022:

build.cmd simlibs_debug

4. Generate Project Files

On Linux:

./blocks_genprojfiles_vscode.sh

On Windows:

blocks_genprojfiles_vscode.bat

Open the generated workspace and launch the Unreal Editor in DebugGame mode.

Headless Execution

To run a packaged Unreal environment with off-screen rendering:

Blocks{.exe/.sh} -RenderOffScreen

To disable rendering completely:

Blocks{.exe/.sh} -nullrhi

See headless and cloud execution for additional configuration details.

Open Source and Professional Services

Project AirSim’s simulation core, APIs, configuration system, extension points, and reference workflows are available publicly under the MIT License. We want the open-source project to be useful for evaluation, research, development, and real autonomy workflows.

IAMAI Consulting Corp. maintains and extends the Project AirSim ecosystem. Drawing on the team’s work on Microsoft AirSim and Project AirSim, IAMAI provides professional services for organizations that need to turn a prototype into a repeatable simulation or validation workflow.

IAMAI can help with:

  • simulation-readiness and architecture assessments;

  • PX4, ROS 2, JSBSim, vehicle, sensor, and autonomy-stack integration;

  • custom Unreal Engine environments and simulation workflows;

  • reproducible scenarios and validated builds; and

  • maintained delivery, updates, and engineering support.

If you have an upcoming integration, demonstration, pilot, or validation milestone, talk to IAMAI about a focused first engagement.

Community and Contributions

Project AirSim grows through practical use, technical feedback, and community contributions.

The roadmap is managed through GitHub issues and discussions. Items labeled roadmap describe planned direction, while items labeled need help identify opportunities for community participation.

Third-Party Interoperability and Licensing

Project AirSim interoperates with third-party engines, libraries, models, and tools, including JSBSim-compatible aircraft definitions. Those components and assets remain under their respective licenses.

Before redistributing or extending an integration, review Project AirSim license information, NOTICE.txt, and the licenses under thirdparty/Licenses/.

Copyright (C) Microsoft Corporation.
Copyright (C) 2025-2026 IAMAI CONSULTING CORP

MIT License

Introduction

Getting Started

Configuration

Physics

Sensors

Controllers

Scene & Simulation

APIs & Programmatic Control

Python Client API Reference