500 Million Pokémon Go Players Unknowingly Built the World's Largest Visual Navigation Dataset for Robots

Niantic has revealed that eight years of AR scans from Pokémon Go's massive player base produced over 30 billion geo-tagged images — now being used to train robots that can navigate within centimeters without GPS.

For eight years, hundreds of millions of people walked their neighborhoods, parks, and city streets pointing their phone cameras at the world around them. They thought they were catching Pokémon. They were actually building one of the most valuable spatial datasets in AI history.

Niantic has disclosed that Pokémon Go's 500-million-plus player base generated over 30 billion geo-tagged images through the game's AR scanning features, as first surfaced by @markgadala in a post that exploded to 5.7 million views. The dataset — accumulated since the game's 2016 launch — now powers a live 3D model of the physical world precise enough to guide autonomous robots to centimeter-level accuracy without relying on GPS.

Get our free daily newsletter

Get this article free — plus the lead story every day — delivered to your inbox.

Want every article and the full archive? Upgrade anytime.

No spam. Unsubscribe anytime.