Microsoft unveiled Project AirSim, offering a realistic platform to build, train, and test autonomous aircraft. The high-fidelity simulation platform was launched at the Farnborough International Airshow and is currently available in a limited preview.
Female aerospace engineer monitors flight simulator.
Capabilities of the Project AirSim Simulation Platform
Project AirSim offers a realistic setting for AI models to process millions of flights quickly. AI models employ these simulations to train how to react to a wide range of potential real-world conditions, such as the weather, temperature changes, or how wind can affect battery life.
Microsoft Azure powers the platform. It produces a lot of data to train AI models on what to do during flight, from takeoff to cruising to landing. Project AirSim has libraries of different simulated 3D environments and customizable, pretrained AI building blocks to speed up autonomy.
The pretrained AI building components feature advanced models for identifying and avoiding obstacles and conducting precision landings. Because of its out-of-the-box capabilities, a person without deep learning knowledge can train an autonomous aircraft with the platform.
Balinder Malhi, engineering lead for Project AirSim, said that they established Project AirSim in the hope that it would help democratize and expedite aerial autonomy through its key capabilities such as capturing and processing massive amounts of data, simulating the real world accurately, and encoding autonomy without the need for deep AI expertise.
Project AirSim might aid in the certification of autonomous systems by creating situations that an autonomous vehicle must successfully navigate. Yet, Microsoft expects the platform to have applications beyond developing AI models.
In 2017, Microsoft Research launched Project AirSim as an open-source project. The open-source tool has been transformed into the Project AirSim end-to-end platform.
ALSO READ: Microsoft Flight Simulator to Return After 13 Years
Project AirSim Users
Airtonomy, a North Dakota-based startup, has been training autonomous aerial vehicles to examine wind farms, scan animals, and find leaks in oil tanks using Project AirSim simulations.
Josh Riedy, Airtonomy CEO, said they needed something other than the physical world to design their solutions for customers. He pointed out that winter in North Dakota can last up to seven months. "You don't want to fly drones into wind turbines, powerlines, or really anything for that matter." Josh Riedy said.
Prior to taking to the skies in the real world, Matt Holvey, director of intelligent systems at Bell, claimed that AirSim gave his team a realistic idea of what to expect. It will be one of the techniques used to speed up scaling aerial mobility. Holvey said that testing and validating everything by hand, in a physical lab, or even by flying an aircraft would take decades, and it would cost billions. He said that Project AirSim advances that via high-fidelity simulation.
Bell has also utilized Project AirSim to refine its drone's capacity to land on its own, monitoring wind farms and surveying wildlife. Bell could quickly train its AI model on tens of thousands of different landing scenarios using Project AirSim.
RELATED ARTICLE: Hypersonic Flight Race: NASA, ANL Develops Aircraft Engine Simulation Using AI and Machine Learning
Check out more news and information on Technology in Science Times.
