airsim

AirSim is an open-source, cross-platform simulator for drones, cars and other vehicles, developed by Microsoft AI & Research. It is implemented primarily as an Unreal Engine plugin that can be dropped into Unreal environments, with an experimental Unity plugin release also available. Key capabilities described in the README: - Simulation fidelity: physically and visually realistic simulation, with software-in-the-loop support for popular flight controllers such as PX4 and ArduPilot, and hardware-in-the-loop support with PX4. - Programmatic control: APIs exposed over RPC and accessible from C++, Python, C# and Java. These APIs allow retrieving images, reading vehicle state, and controlling vehicles. The same APIs are available as a separate cross-platform library so code can be tested in simulation and later run on real vehicles. - Data collection: training data can be recorded via an in-simulator record button or through APIs for full control over what, where and when data is logged. - Computer Vision mode: a mode without vehicles or physics where cameras can be positioned arbitrarily to collect depth, disparity, surface normals or object segmentation images. - Weather effects: configurable weather options accessible via keyboard (F10) and APIs. - Manual control: drones can be flown with an RC controller; cars can be driven with arrow keys. Getting started: precompiled binaries are offered for Windows and Linux, with build instructions for Windows, Linux and macOS. Detailed documentation covers build, APIs, settings, image APIs, reinforcement learning and design. The project targets AI research, including deep learning, computer vision and reinforcement learning for autonomous vehicles. It is released under the MIT License and has an associated paper from the FSR 2017 conference. Community channels include GitHub Discussions and a Facebook group.