Friday, September 22, 2017

Microsoft Trains Unmanned Gliders To Make Decisions On Flight

Microsoft Trains Unmanned Gliders To Make Decisions On Flight
The self-driving cars are yet to go mainstream as there’s still a lot to be taken care of. All the major motor companies like Tesla, Ford, Nissan, BMW and Bosch have been working on launching these autonomous vehicles for quite some time now. Tesla became the first motor company to venture into this futuristic project followed by the others. Google is also working on the technology for several years. By the year 2020, there’s a sharp possibility that many self-driving cars will hit the roads.

Microsoft however, is trying something new this time. The company is building autonomous gliders with self-decisive capabilities. The project is initiated by Microsoft under the supervision of Ashish Kapoor, a computer engineer who happens to be the Principal Researcher at Microsoft. The mission is to train the autonomous gliders to take important decision on the fly. Kapoor and his team had recently conducted a flight test in the American state of Nevada as reported by New York Times.


The Problem Statement
As revealed by Microsoft, they are hoping that by implementing advanced AI algorithms, the gliders can be made smart enough to re-route themselves in the hour of adversity. For instance, if some area is affected due to bad weather conditions, they must be able to detect that earlier and change their direction accordingly. This will further boost their life-cycle and allow them to stay aloft for a longer duration. The project is led by Ashish Kapoor who is an AI researcher and a licensed pilot.

The Microsoft team is working on smart AI algorithms for detecting certain entities like air temperature, weather conditions and direction of the wind to decide which path should be followed by the drones and gliders. To make sure that the drones are able to make such decisions on their own, Microsoft is using a highly advanced AI technique called Markov Decision Process. It’s a mathematical framework that allows computers to make decisions on their own. The outcome however, is partly under the control of a third party.


No comments:

Post a Comment