Google explains how it predicts traffic, plans routes on Maps

Google Maps uses machine learning in combination with various data sources including aggregate location data, historical traffic patterns, local government data, and real-time feedback from users, to predict traffic.

Google explains how it predicts traffic and plans routes on Maps.

(Subscribe to our Today's Cache newsletter for a quick snapshot of top 5 tech stories. Click here to subscribe for free.)

Google Maps is used by numerous people on a daily basis while traveling as the navigation platform effectively predicts traffic and plots routes for them.

The search giant says, over one billion kilometres are driven with Google Maps every day in more than 220 countries and territories around the world.

Google Maps uses machine learning in combination with various data sources including aggregate location data, historical traffic patterns, local government data, and real-time feedback from users, to predict traffic.

Google says it has been working with DeepMind, an Alphabet-owned AI research lab, to improve the accuracy of its traffic prediction capabilities. By using Graph Neural Networks (GNN), a machine learning architecture, it has managed to reduce the percentage of inaccurate expected time of arrivals (ETAs) even further.

Google Maps ETA improvements around the world.

 

“This technique is what enables Google Maps to better predict whether or not you’ll be affected by a slowdown that may not have even started yet,” Johann Lau, Product Manager, Google Maps, said in blog post.

Also Read

Google adds more colours and details in Maps app

 

The researchers at DeepMind have divided the road network into “Supersegments.” These are smaller groups of adjacent roads that share traffic volume. The components that drive the prediction system include a route analyser and a GNN model.

The route analyser processes terabytes of traffic information to construct Supersegments, while the GNN model is optimised with multiple objectives and predicts the travel time for each Supersegment, the researchers explained.

Supersegments consists of multiple adjacent segments of road that share traffic volume.

 

When plotting routes, Google Maps also uses the predictive traffic models to assess the routes that are likely to receive heavy traffic, and accordingly suggests alternative routes with lower traffic. It also takes into account other factors such as quality, size and direction of the road.

Also Read

Google Maps launches end-to-end biking directions with bikeshare information

 

“Since the start of the COVID-19 pandemic, traffic patterns around the globe have shifted dramatically,” Lau said. “We saw up to a 50% decrease in worldwide traffic when lockdowns started in early 2020.”

The predictive traffic models have been updated to work with the decrease in traffic flow. Now, the models automatically take into account historical traffic patterns from the last two to four weeks, over patterns from any time before that, Google said.