Pokémon Go Data Powers Centimeter-Accurate Delivery Robots

Data generated by the augmented reality game Pokémon Go is now being utilized to enhance the navigation capabilities of delivery robots, thanks to advancements from Niantic Spatial, an AI spinout formed in 2025. This technology is enabling more reliable and precise deliveries in challenging urban environments.

From Virtual Creatures to Real-World Logistics

Pokémon Go, released in 2016 by Niantic, quickly became a global phenomenon, attracting hundreds of millions of players to explore real-world locations. According to Scopely, the game still had over 100 million players in 2024. This massive player base inadvertently created a vast dataset of images tagged with precise location information. Niantic Spatial is now leveraging this data to build a high-precision world model.

Niantic Spatial’s Visual Positioning System

Niantic Spatial has developed a visual positioning system (VPS) that can pinpoint a device’s location to within a few centimeters using only camera input and map context. The system was trained on approximately 30 billion images collected from Pokémon Go and Ingress players, providing a dense global dataset with metadata including latitude, longitude, camera orientation, and device pose.

Coco Robotics: A Real-World Application

Coco Robotics, a last-mile delivery startup, is utilizing Niantic Spatial’s VPS to guide its fleet of sidewalk robots through dense urban areas where GPS signals are often unreliable. The robots, which can carry up to eight XXL pizzas or four shopping bags, operate in cities like Los Angeles, Chicago, Jersey City, Miami, and Helsinki, having completed over half a million deliveries.

The Challenge of Urban Navigation

GPS signals often struggle in urban “canyons” created by tall buildings, leading to inaccuracies. Niantic Spatial’s VPS addresses this issue by relying on visual cues rather than radio signals, allowing robots to maintain accurate positioning even in areas with poor GPS reception.

Beyond Delivery: The “Living Map” Concept

Niantic Spatial envisions its technology as a foundation for a “living map”—a dynamic, hyper-detailed virtual simulation of the world that continuously updates with real-world changes. As robots and other devices gather more data, this map will become increasingly accurate and comprehensive, benefiting a wide range of applications beyond delivery services.

Implications for Robotics and AI

The success of this partnership highlights a shift in the primary consumers of mapping technology – increasingly, it is machines rather than humans. This technology could be crucial for enabling robots to safely and effectively navigate complex environments, paving the way for wider adoption of robotic solutions in various industries.