Google announces improvements to the arrival time calculation of its Maps navigation. The company now says it can make more accurate predictions using artificial intelligence.
Google Maps arrival time
Users of Google Maps all put together annually a billion kilometers in 200 countries worldwide. With navigation, Google also always shows the expected arrival time of a route, but often traffic congestion can throw a spanner in the works. That’s why the Google Maps team partnered with its DeepMind artificial intelligence (AI) lab to make better arrival time predictions. Google announced this in a blog post.
Google Maps previously calculated traffic density by taking into account a combination of two factors: current traffic conditions and historical traffic patterns. Nevertheless, it became very clear to Google this year that this method is not ideal, because due to the consequences of the coronavirus, the company saw the traffic during rush hour decrease by 50% due to the lockdowns at the beginning of this year.
This meant that the predictions of the expected arrival time were a lot less accurate and Google was forced to respond immediately to the changed situation. The company did this by looking at historical traffic patterns in a shorter-term from now on and not looking too far into the past.
Smart algorithms
Still, Google said it went a step further in predicting traffic because DeepMind has developed a graphical neural network for this. First of all, DeepMind divides the traffic in a city into different clusters of streets that have about the same degree of congestion.
Google then makes a predictive algorithm work per cluster, and in addition, it also allows the clusters to dynamically change shape as the traffic volume also changes. Each predictive algorithm also takes a variety of factors into account, including the time of year, road quality, speed limits, accidents, and road diversions.
Google is also already sharing the results of the AI predictions because in some cities the company was able to make the predictions by half more accurate. Google is also experimenting with paying parking fees in Maps, there will be a dark theme for the app, and maps will be more detailed. What do you think of the estimated arrival time in Google Maps? Do you think Maps does better than any other navigation app? Let us know in the comments at the bottom of this article.