Microsoft AIRO group on January 24, 2017.(Photography by Scott Eklund/Red Box Pictures)

Virtual world are suddenly in high demand. Considering that we have been creating a large number of technologies that have yet to be tested properly in real conditions, they have come up as the best alternative to avoid causing some serious damage during the course of conducting the tests. Along the same, Microsoft has open sourced an advanced virtual world in beta.

The software has been published on GitHub and can mimic the real world with a high degree of accuracy — apparently to the point of creating things like shadows and reflections. Microsoft hopes that by open sourcing this world, it will be able to give developers and researchers an avenue of testing their technologies without causing any real or lasting damage.

The advantages? Let’s explore them with a simple example, say that of a self driving car, since they are the hottest test-hungry technology now a days. Say you have written code for a self-driving car and want to test it. On e of the options available to you is to convince a manufacturer to actually create a custom made system that would support your code, get the government to sanction testing of the loaded system on public roads — with the risk that it could go berserk and end up causing damage. Not very appealing.

The other road available to you, is to load the code into a car of your choice in a virtual world, and let it cruise through the roads. You can let it drive into homes or drown into a rive for all the real damage that will be caused. Meanwhile, all the hassles associated with getting a real car, making it compatible with your system, obtaining government permissions and so on and so forth are avoided. Once you have perfected your system, then and only then you can take things further.

There are other benefits to this system as well. Say you were trying to teach an autonomous drone that it shouldn’t run into building, that buildings are bad. A mistake there would cost you an undoubtedly expensive drone, and force you to start the work over. Load the code into a virtual drone and let it bang away at walls until it learns better. And of course, you can increase the pace of the learning so that the systems run through scenarios faster and learn faster.

Of course, there are a few advantages associated with real-world testing as well. For example, this would work if the virtual world in question was a highly accurate representation of the real world. So, a simulator where the drone can go through humans wouldn’t do. Why? Because if you were to load that code into a real drone, you would have it attempt to go through people because from its perspective, it should be possible.

However, that is the only dark spot on an otherwise bright horizon and with proper care, can be avoided completely. Microsoft’s simulator can be used to test drones, self-driving cars, anything that requires teaching robots their way about the world. The platform could also help developers inoculate something resembling the ability to predict, that is possessed by humans.

The Aerial Informatics and Robotics Platform could help developers make advances in planning capabilities, which aim to help the gadgets anticipate what will happen next and how they should respond, much like humans know to anticipate that cars will drive by when we cross a street. That kind of artificial intelligence – which would closely mimic how people navigate the real world – is key to building practical systems for safe everyday use.

So basically, Grand Theft Auto just got real — and useful.

Leave a Reply

Your email address will not be published. Required fields are marked *