Pokémon Go, a game that was once known for sending players outdoors to capture virtual creatures, is now contributing to real-world robotics. The game’s data is helping robotics companies to train delivery systems across cities in the US and other countries. Over the past decade, players have voluntarily submitted photos and short videos of public landmarks, street corners, storefronts, and urban intersections, creating a dataset of around 30 billion ground-level images across major cities worldwide.
Niantic Spatial, the enterprise AI and mapping division spun off from Pokémon Go developer Niantic Inc., has been converting this crowdsourced data into a photorealistic, continuously updated street-level model designed for robotic navigation.
This model is currently being used by Coco Robotics to operate a fleet of roughly 1,000 delivery bots in cities including Los Angeles, Chicago, Miami, Jersey City, and Helsinki, where the robots have logged millions of miles of deliveries to date. Brian McClendon, Niantic Spatial’s chief technology officer and one of the original creators of Google Earth, explains how the data strategy works.
What Pokemon Go makers said about the game’s data helping robotics companies
In a statement to Fortune, McClendon said, “We look at the player data as very high-quality ground training data for other lower-quality datasets.
The long-term philosophy of Niantic Spatial is that we can solve these hard problems of localization, reconstruction, and semantics by using very concentrated places to train models and then use much more broadly available data at lower resolution to be able to localise, visualise, and understand from ‘bad’ data.”
The company’s dataset, built from around 30 billion Pokémon Go images, is being used to develop a real-world, real-time mapping system. These player-generated scans help train models to recognise precision and identify inconsistencies in input data. The approach reflects Niantic Spatial’s shift towards building mapping and spatial intelligence systems using user-generated data.
Niantic Spatial’s Visual Positioning System (VPS) is designed to address limitations of GPS in dense urban areas, where tall buildings can disrupt satellite signals. Instead of relying on satellites, VPS compares live camera feeds with its image database to determine a device or robot’s position in real time, which can be useful for applications such as autonomous deliveries.
“The model will work in real time, taking in images from the robot and comparing them to both publicly available as well as proprietary datasets we’ve collected to determine the robot’s global position and heading,” a Niantic Spatial spokesperson told Fortune.
The spokesperson added, “Niantic Spatial’s VPS is particularly resilient in urban canyons where GPS performs badly.”
The company said its early VPS models were trained using scans voluntarily submitted by players. “Our initial VPS was built using scans that users choose to take in games—but no single source defines the model,” the spokesperson noted.
Over time, the system has expanded to include data generated by enterprise users. The underlying large geospatial model is trained on billions of images and scans, enabling 3D reconstruction, localisation, and semantic understanding of environments.
As CEO, John Hanke wrote, “For the past several years, we’ve been building a large geospatial model that acts as a living, breathing map of the world, one that is native to robots and AI.”
Coco CEO Zach Rash pointed to limitations in current robotic navigation systems. She told Fortune, “Robots don’t have the same intuition yet as a human, where a human can understand, ‘My GPS isn’t really working, but I understand that’s probably the right place to go.’ We need the robot to have that sort of intuition.”
He added that VPS can help in dense city environments. “When we go into really dense areas with high rises, that’s where the VPS solution can be really helpful. Our GPS and our existing solutions might fail in that sort of environment,” Rash explained.
Highlighting the impact on users, Rash said, “It is a terrible customer experience if the robot parks in the wrong place waiting to receive that order.”
“It’s very early with [Niantic Spatial], and I think we’re excited to collaborate with such an incredible team on figuring out how we add this toward existing technology to make the service better. VPS is an obvious one. They’re very good at doing this. If I can more precisely figure out where to drop off food, my customers will be happy,” he continued.
