Microsoft Crash-Test Drones, Robots, and Self-Driving Cars Simulator AirSim available Open Source Now. One if the biggest challenge faced while designing a self-driving vehicle is the irregularities of the outside surroundings. From electricity poles to bumpy road anything can become an obstacle for your creation. Keeping this issue in mind, Microsoft has designed and open-sourced a simulation course that will train self-driving cars, robots, and drones so that they learn to avoid obstacles in real life.
It is named the Aerial Informatics and Robotics Platform, or AirSim in short. It is available for Linux and Windows. It will let you create a casual environment for you to train your vehicle and experiment with various models and terrains.
It should be mentioned here that this invention isn’t made with the aim of totally eradicating real world testing, on the contrary, it is made to compliment it. It will allow you to create many situations and as many times as you want to.
In addition to creating various real world scenarios, the AirSim will let the user incorporate lifelike physics systems to precisely simulate many lighting conditions and object groups (such as a bunch of trees) to train robotic models in understanding how to evaluate depth, avoid hurdles and deal with shadows and glare while navigating through outdoor spaces.
While AirSim is designed to support controls for DJI and MavLink-based drones out of the box, it is also being designed to eventually work with other self-driving vehicles, hardware platforms, and software protocols by a simple method of just adding modules to the software.
Let us know how you feel about this new technology and how you feel it will revolutionize the world of self-driving machines in the comments.