If you have an Android mobile with Google services, you will have Google Maps installed by default as a navigation application. Apart from having a huge database with maps and routes, Maps shows us traffic information in real time.
In a new blog post, Google wanted to explain how are they able to predict traffic and establish the best routes. Machine learning, gathering user data, and more – this is how Maps can analyze traffic.
13 TRICKS to get the MOST out of GOOGLE MAPS
This is how Google Maps determines the traffic state
When we open Google Maps and establish a route, the application It tells us the estimated travel time, the path to take, if there will be traffic or not (and in which sections) and, ultimately, all the necessary information related to said route. Google Maps uses information from the location data you get from users, to be able to report in real time on the traffic estimates.
Google does not only use user data in real time, it has a history of the behavior of the roads to make predictions of the traffic state
However, from Google they indicate that this information is not enough to predict what will happen after 10, 20 or 50 minutes of travel, and that this is where “technology really comes into play.” To predict how traffic will behave in certain periods of time, Google uses machine learning. It does this by combining historical road traffic patterns over time. That is, Google knows that on X roads, between certain hours, X vehicles usually travel at a certain speed.
Google makes it clear that, when selecting the route, it also takes into account the type of road. Apart from knowing if the route is paved or if it is dirt, they know the size of the road and how frequent it is that it stops more in one or the other, which also helps to predict the routes.
They also use authorized data from local governments to know the speed limits, tolls or to know if some roads are restricted to traffic. This information is combined with that of the users, which is what really allows us to know if there is any work, closed or collapsed lane, objects on the road and others.
Google also uses DeepMind technology to improve prediction capabilities
Similarly, Google uses technology from DeepMind, an AI lab, to improve prediction capabilities. According to the company, currently Google can predict traffic conditions with 97% prediction.
They also say that, since the beginning of the COVID-19 pandemic, they noticed up to 50% decrease in global traffic and that machine learning models have been updated with the latest data to maintain prediction accuracy.
Via | Google