Artificially intelligent monitoring systems can be immensely helpful in keeping overall costs down, but AT&T Labs is looking into ways to make them even better through the addition of modern drone technology. According to Art Pregler, the unmanned areal systems (UAS) Program Director at AT&T, the company wants to make use of machine learning and associated algorithms to keep cell towers and other infrastructure in prime condition. The way it would work is simple enough. A drone could, in theory, fly out to cell sites and fly around the equipment, analyzing for any defects or degradation. If any problems are detected during the fly-by, the drone can call in a crew or take other actions as needed to fix those issues.
In fact, Pregler says the drones currently under development have already learned to detect the equipment itself. So the only thing left to do is teach them to identify any abnormalities or problems in the equipment and how to react. The team at AT&T Labs ended up using more than 6,000 videos and over a million cell tower photos to teach the drones to accomplish that task alone. There is still a lot more work to be done, according to Pregler, who says that they must be taught to identify "over 200 types of elements on a cell tower." So this is definitely still in its early stages.
The obvious benefit to an A.I.-based drone system is that it would keep infrastructure at a more cost-effective level of maintenance. As is often the case with things that degrade over time, and as things mostly stand right now, problems don't always become clear until after the damage has already been done to the network itself. In other cases, problems don't present themselves until beyond a point at which it can be simply repaired – leading to added cost associated with replacing infrastructure. There is an additional cost associated with personnel, as well, if a company wants to maintain a watch on that infrastructure. Since the company has already been putting a lot of research into other uses for drone technology for quite some time, AT&T is in a very good position to find out if an A.I. solution is really viable commercially. However, AT&T is also already currently facing internal problems due to worker treatment, so it's probably a good thing that this maintenance drone project is still in its infancy.