Microsoft’s AirSim (Aerial Informatics and Robotics) is an open source artificial intelligence simulator. The technology was first made available to designers of drones in order to assess safety features. The origins of the technology were for the game development tool suite Unreal Engine. The platform enables interfaces to common robotic platforms such as a Robot Operating System (ROS).
Now the technology has been expanded to provide realistic environment simulations that enable research into how autonomous vehicles will behave and operate in the open world.
Open-sourced simulations
The platform remains an open, community-driven system designed for testing algorithms. The platform, Smart2Zero reports, provides a detailed three-dimensional urban environment. Here conditions can be varied to allow safety features to be fully tested. This includes having operating traffic lights; aspects of the urban landscape like parks, lakes, and construction sites.
A simulation full of options
Developers will also be able to put their systems to the test in a variety of neighborhoods, ranging from downtown, semi-urban, countryside, and built-up industrial environments. There are some 12 kilometers of roads and 20 city blocks.
Moreover, as the car simulation is separate from the environment that it operates in, developers are able to construct an environment appropriate to the vehicle being tested. The functionality of AirSim further allows users to add things like new sensors or vehicles into the final simulation.
AirSim’s application programming interfaces can be deployed in different languages. This means it can be used in conjunction with different machine learning tools. AirSim will also function with Microsoft Azure (a cloud computing service created by Microsoft for building, testing, deploying, and managing applications and services). This is to scale up the training for machine learning algorithms.
Microsoft plans to expand AirSim further with different sensors, better vehicle physics, weather modeling, and more additional realistic environments.
For more on autonomous car technology, read the Digital Journal article “New high resolution LiDAR becomes more affordable.”
