Microsoft AirSim: One of the biggest challenges in building autonomous vehicles is dealing with the irregularities and harsh conditions of outdoor environments: From battery life to how they react on winding roads and uneven terrain, flying in wind and rain, there are many things that experts will have to control in their creations.
With that in mind, a team of Microsoft researchers has developed an open source simulation tool to help train and control autonomous vehicles, drones, and robots. The platform can “teach” the vehicles to avoid obstacles just as they would in real-world conditions.
Called the Aerial Informatics and Robotics Platform, or AirSim for short, the application is available for Linux and Windows, and will allow you to create a randomized environment to train your bot or vehicle, experiment with different models, and conduct tests in various scenarios and environments.
It is worth noting that the software is not designed to entirely replace real-world testing, but rather to complement it, allowing you to simulate a wide range of tests as many times as necessary.
In addition to creating situations that mimic the real world, AirSim also includes systems for realistic simulation in various lighting conditions and tree stands, which will help train robotic models to understand how to calculate depth, avoid obstacles, and so on.
AirSim supports controls for DJI and MavLink drones out of the box, but also supports other autonomous vehicles, platforms and software protocols by adding special modules to the software.
If you want to try AirSim you can download it from its GitHub repository.
https://github.com/Microsoft/AirSim
