The company Airtonomy, based in North Dakota (USA), uses the AirSim project to train autonomous aerial vehicles that inspect critical infrastructure such as wind turbines. Image: Airtonomy.
Several years ago, Microsoft Research offered an open access tool called AirSim. It was a simulator for drones, cars and other autonomous vehicles. After years of experimentation, Microsoft ended up archiving the original AirSim platform repository. But this week, the company launched a new AirSim simulation platform for the aerospace industry, called “Project AirSim”.
The AirSim project, like the original AirSim, uses artificial intelligence models to build, train, and test drones and other autonomous aircraft using 3D simulation. According to the officials, one of the main differences between the original AirSim system and the new project is that the former required “deep expertise in coding and machine learning”.
The first AirSim was integrated with Bonsai machine learning technology, as well as reinforcement learning from Azure Machine Learning Services.
From amateur drone to transport aircraft
The new AirSim project will offer libraries of simulated 3D environments. It will also use Azure to generate “massive amounts of data to train AI models on the exact actions to take in each phase of flight, from takeoff to cruise to landing,” according to Microsoft.
The software giant foresees a number of potential use cases for the AirSim project, ranging from hobbyist drones to passenger planes. Autonomous aerial vehicles can be used to inspect equipment, transport packages and more.
The AirSim project is currently in limited preview. For more information, fill out this form.
Source: ZDNet.com