If you own an Android phone, chances are that you’ve got Google Maps services preinstalled once you’ve got it. This tool one of Google’s most popularly used applications amongst its suite of apps for its proven useful capabilities and features. One of its key features, by providing estimated time arrival (ETA) data to its users are amongst the sought-after functions most users rely on when they are on the app.
In a report by James Vincent on The Verge, Google shared that it used the help of its partner artificial intelligence (AI) arm, DeepMind, to be able to help its features attain a high level of accuracy.
But how did they do it? According to Vincent, Google and DeepMind took data from several sources including but not limited to live traffic data taken in confidentiality amongst Android users, traffic data from the past, road quality, as well as publicly available information such as ongoing constructions and certain speed limits on certain roads and streets.
This large swath of information is then fed to the AI tool which then makes its calculations in order to help generate future possible patterns in traffic. The search giant also declares that with the help of AI and all of the information collection that was conducted, Google Maps’ services were able to register up to 50% increase in accuracy of its ETA features in cities all over the globe.
To be able to deliver such improvement in its mapping accuracy levels, it is reported that Google used a system called “supersegments” to be able to identify and analyze the data coming from these areas. These are clusters or groups of nearby streets which share a similar volume of traffic. Then a neural network in AI known as Graph Neural Network is used to be able to analyze and process the data being collected and fed to the system.
In the part of the lowered traffic and number of cars going to and fro due to the Coronavirus disease – 2019 (COVID-19) pandemic, the company shared that it changed how it gathered data in order to still come up with correct and efficient findings for the benefit of its users.