Microsoft has released a new open source simulator which can crash-test drones, robots and self-driving cars virtually, before testing them in the real world.
The company in a blog noted that drones, autonomous vehicles and robots are still in the early stages of development and at present cannot differentiate obstacles such as people, trees and signal posts from things such as shadows, reflections and clouds.
To help developers of drones, robots and autonomous vehicles, Microsoft is releasing a simulator software, where they can be trained and tested, so that they can operate autonomously without running into any obstacles in the real world.
This is part of Microsoft’s research project called the Aerial Informatics and Robotics Platform, which includes software that allows researchers to write code that controls aerial robots and other gadgets in a realistic simulator.
According to Microsoft, the tools can help in creating artificial intelligence system that can drive cars and deliver packages.
Microsoft researcher and project lead Ashish Kapoor said: “The aspirational goal is really to build systems that can operate in the real world.”
He said: “That’s the next leap in AI, really thinking about real-world systems.”
Until recently, simulators offered limited help as it was difficult to visualise real-world complexities in the virtual world. But, thanks to improvements in graphics hardware, computing power and algorithms, the new simulator has been designed to offer a realistic view of the environment.
Microsoft claims that the simulator is built on photorealistic technologies that can accurately render things such as shadows and reflections with high precision. This can make a significant difference in computer vision algorithms.
Microsoft principal research software development engineer Shital Shah said: “If you really want to do this high-fidelity perception work, you have to render the scene in very realistic detail – you have sun shining in your eyes, water on the street.”
As the new simulator is realistic, but not actually real, researchers can, in turn, use it as a safe, reliable and cheap testing ground for autonomous systems.
This has two main advantages. First, it means they can “crash” a costly drone, robot or other gadget an infinite number of times without burning through tens of thousands of dollars in equipment, damaging actual buildings or hurting someone.
Second, it allows researchers to do better AI research faster. That includes gathering training data, which is used to build algorithms that can teach systems to react safely, and conducting the kind of AI research that requires lots of trial and error, such as reinforcement learning.