Innovation

Microsoft's AI system for training autonomous cars and drones goes open source

Microsoft's AirSim is a platform for researchers to work with technologies like deep learning and computer vision to improve the operation of autonomous vehicles.

Want to use artificial intelligence (AI) to train autonomous drones and self-driving cars the way Microsoft does? Well, now you can. Microsoft recently shared its open source system for performing such trainings, including its AI-based platform AirSim.

The project, called the Aerial Informatics and Robotics Platform, aims to provide simulation tools and resources to help in the training of robots, drones, and other hardware that can operate autonomously, according to a Microsoft blog post. By improving the way that these devices and vehicles can navigate their environment, the hope is that that can eventually shed their "emerging technology" title.

By using virtual simulators, like AirSim, a team is able to collect valuable data and train the systems more efficiently before they take them out of the lab. "The aspirational goal is really to build systems that can operate in the real world," Ashish Kapoor, a Microsoft researcher, said in the post.

SEE: Gallery: 10 self-flying auto-follow drones to take on your weekend adventures

The problem with previous simulators is that they weren't accurate enough to properly train the systems, the post said. However, graphics advances and photorealistic technologies have improved the simulation, and have also improved the data that can be captured by computer vision programs.

"The biggest advantage of this simulator is that it uses recent advances in computation and graphics to simulate the physics and perception such that the environment realistically reflects the actual world. Such realism can enable efficient training and testing of machine learned models by generating vast quantity of ground truth data," the research abstract said.

AirSim was built on the Unreal Engine, and is available on GitHub through an open source license. The technology is also cross platform, and "supports hardware-in-loop with popular flight controllers such as Pixhawk for physically and visually realistic simulations. It is developed as an Unreal plugin that can simply be dropped [into] any Unreal environment you want," the GitHub page noted.

The Aerial Informatics and Robotics Platform also provides a software library that assists developers in writing code for drones built on either the MavLink or DJI platform. Any autonomous technology that seeks to navigate its surrounding can work with this platform, the post noted.

"I can actually use the same code base to fly a glider or drive a car," Kapoor said in the post.

In open sourcing the project, the team hopes to help democratize the advancement of robotic and autonomous technologies. According to the post, Microsoft is hoping that it will help speed up the development of standard protocols and regulations as well.

The 3 big takeaways for TechRepublic readers

  1. Microsoft has shared information on its open source Aerial Informatics and Robotics Platform for developing and advancing autonomous systems.
  2. An AI-based simulator, AirSim, helps research teams better test autonomous drones with more photorealistic virtual environments that produce better test data.
  3. The platform also offers a software library to help developers more quickly build out these autonomous systems and get to work testing them.

Also see

About Conner Forrest

Conner Forrest is a Senior Editor for TechRepublic. He covers enterprise technology and is interested in the convergence of tech and culture.

Editor's Picks

Free Newsletters, In your Inbox