Microsoft Offers to Train Autonomous Drones in Virtual Worlds

pcmag.com:

Microsoft is offering to train autonomous drones before their first real flight by using millions of practice flights and scenarios in a virtual world.

The platform is called Project AirSim(Opens in a new window) and it runs on Micorsoft Azure. It offers anyone who needs to develop a new, autonomous drone, the option of using realistic environments, AI models, and deep learning to run through millions of flights in seconds. By doing so, every possible scenario a drone will face can be pre-tested before the first real flight.

Microsoft views AirSim as being used to pre-train drones for very specific tasks such as aerial infrastructure inspection, last-mile delivery, and urban air mobility. And all the training can be performed in a virtual environment that matches the real-world version it will end up working in.

"Autonomous systems will transform many industries and enable many aerial scenarios, from the last-mile delivery of goods in congested cities to the inspection of downed power lines from 1,000 miles away," said Gurdeep Pall(Opens in a new window), Microsoft corporate vice president for Business Incubations in Technology & Research. "But first we must safely train these systems in a realistic, virtualized world. Project AirSim is a critical tool that lets us bridge the world of bits and the world of atoms, and it shows the power of the industrial metaverse – the virtual worlds where businesses will build, test and hone solutions and then bring them into the real world."

Project AirSim started as a Microsoft Research project back in 2017(Opens in a new window) and used both the Unreal Engine and Unity to produce the realistic 3D virtual environments for the drones to fly around. It proved very successful over the