Google has joined hands with DeepMind to develop a new Machine Learning (ML) algorithm for accurate traffic predictions and estimated times of arrival (ETAs).
Google Maps’ ETA predictions already have a very high accuracy bar. According to Johann Lau, Product Manager, Google Maps, they have been consistently accurate for over 97% of trips.
By partnering with DeepMind, Google has been able to cut the percentage of inaccurate ETAs even further by using an ML architecture known as Graph Neural Networks.
The team could improve the accuracy of real time ETAs by up to 50% in places like Berlin, Jakarta, São Paulo, Sydney, Tokyo, and Washington D.C
Since the start of the COVID-19 pandemic, traffic patterns around the globe have shifted dramatically. Google saw up to a 50 percent decrease in worldwide traffic when lockdowns started in early 2020.
Since then, parts of the world have reopened gradually, while others maintain restrictions. Google recently updated its models to become more agile to account for this sudden change. It automatically prioritized historical traffic patterns from the last two to four weeks, and deprioritized patterns from any time before that.
Google’s predictive traffic models are also a key part of how Google Maps determines driving routes.