It’s no secret that Google Maps is hands down the best free mapping service, and it performs noticeably better than its competition under all sorts of conditions – but rarely do people stop to think about the technological implications of a system like this. Even the more tech-savvy of its users probably don’t realize the extent of Google’s operations, and how much the Maps servers are actually doing.
Not everyone realizes it, but there was actually a somewhat specific time period during which Google Maps became truly good, even though it was still doing okay before that – and that was the introduction of Street View. While the implications for the end user are obvious, according to people close to the development project, there is actually more to this feature than user convenience.
Apparently, Google are running very advanced visual recognition algorithms on the materials obtained during their Street View trips – and their systems are frighteningly intelligent. According to reports, the system can extract various important details from a picture – divisions of streets, signs, business names and much more. Then, by combining this data from all collected pictures, it can put together a much more accurate database of its streets and general maps.
The “magic” doesn’t end here though – according to Google, they’re in the process of using this information to develop even more data layers for the service, as the algorithms are providing them with more than they can collect on their own. It’s not known if Google are planning on releasing those layers to the public anytime soon, however – and whether they will reveal any of that information at all.
Google have also recently acquired yet another tech company, this time in the field of satellite imaging – Skybox. According to Google, this has been quite helpful in their development of Maps, especially with regards to 3D models of the environment. Google are known for simply acquiring the technology that they need in cases like this, and their deal with Skybox was immediately noticed by people who follow Google Maps closely.
It’s also interesting how all of this could potentially tie into Google’s development of autonomous vehicles. It’s likely that they have borrowed algorithms from that project for the recognition features of Maps, or even the other way around. If the quality of Google’s programming really is that good, when their self-driving cars are finally released one day, they could be a major game changer if they perform that well. There have already been reports that Google have only seen one or two incidents during their regular tests of those cars, and in all cases it’s been the result of another driver’s error.