Pokémon Go Players Are Helping Build Niantic’s AI-Powered 3D Map
Mike Colagrossi
Niantic, the company behind Pokémon Go, is leveraging data from millions of players to create a revolutionary Large Geospatial Model (LGM). This AI-powered system uses user-contributed scans to map the physical world in unprecedented detail, paving the way for advanced augmented reality (AR), urban planning, and logistics. According to Niantic’s blog post, the LGM combines billions of images and 3D scans captured from games like Pokémon Go and Ingress.
Niantic’s Chief Scientist, Victor Prisacariu, explained, “Using the data our users upload when playing games like Ingress and Pokémon Go, we built high-fidelity 3D maps of the world, which include both 3D geometry… and semantic understanding [of objects].” These maps aren’t just visual—they’re intelligent, capable of inferring unseen details based on patterns and data from similar structures.
The Visual Positioning System (VPS), a feature in Pokémon Go, enables Niantic to collect data with centimeter-level accuracy. With over 10 million locations scanned and a million new scans uploaded weekly, this player-generated content helps Niantic refine its spatial understanding of the world. As the company explains, this data is unique because it’s captured from a pedestrian perspective, covering places inaccessible to cars or traditional mapping methods.
The applications for this technology extend beyond gaming:
- AR Experiences: Players interact with features like Pokémon Playgrounds, where Pokémon can be placed in real-world locations.
- Urban Planning and Logistics: The LGM could assist with navigation, cultural preservation, and spatial analysis of cities.
- Future Technology: Niantic sees the LGM as a foundation for AR glasses and robotics, capable of transforming how humans and machines interact with urban spaces.
Prisacariu notes, “Even the most advanced AI models today struggle to visualize and infer missing parts of a scene… Spatial intelligence is the next frontier of AI models.”
