What Happened
Niantic announced a partnership with Coco Robotics to power delivery robots using data collected from Pokémon Go players over the past eight years. The company’s Visual Positioning System (VPS) relies on more than 30 billion images captured by players who were encouraged to scan real-world statues, landmarks, and buildings in exchange for in-game rewards.
The data collection effort received a significant boost in 2020 when Niantic added “Field Research” features that prompted players to scan physical locations with their cameras. These scans were clustered around strategic locations that served as important spots in Niantic’s games, such as Pokémon battle arenas and PokéStops.
Unlike traditional GPS systems that can drift 50 meters or more in dense urban areas—what experts call “urban canyons”—Niantic’s VPS can pinpoint locations within several centimeters by analyzing visual landmarks captured in the player-generated images.
Why It Matters
This development represents a significant breakthrough in autonomous navigation technology and highlights how gaming companies can leverage player participation to build massive real-world datasets. The partnership demonstrates the commercial value of crowdsourced data collection, even when players may not fully understand how their contributions will be used.
For delivery companies, this technology could revolutionize last-mile logistics by enabling robots to navigate complex urban environments with unprecedented precision. Traditional GPS struggles in cities where tall buildings block satellite signals, causing navigation errors that can put delivery robots on the wrong street or facing the wrong direction.
The privacy implications are equally significant. While players consented to sharing data through the game’s terms of service, many likely didn’t anticipate their casual gaming would contribute to training commercial AI systems for autonomous robots.
Background
Niantic has been collecting location and image data since launching Ingress in 2012, followed by Pokémon Go in 2016. The company positioned itself as an augmented reality pioneer, overlaying digital content onto real-world locations.
The systematic collection of visual data accelerated when Niantic introduced AR scanning features, incentivizing players to capture detailed images of landmarks, statues, and buildings. These “research tasks” rewarded players with in-game items while simultaneously building one of the world’s largest datasets of urban visual information.
Niantic Spatial, the company’s AI division, has been developing the Visual Positioning System to commercialize this data beyond gaming applications. The technology represents a shift from entertainment-focused AR to practical applications in robotics and autonomous systems.
What’s Next
Coco Robotics plans to deploy delivery robots powered by Niantic’s VPS technology for food and grocery delivery services. The robots will use the visual positioning system to navigate sidewalks and urban areas with precision that exceeds traditional GPS-based navigation.
Niantic claims to have precise location data for over one million locations worldwide, positioning the company to expand partnerships with other robotics and autonomous vehicle companies. The success of this deployment could lead to broader adoption of crowdsourced visual datasets for AI training.
The partnership also sets a precedent for how gaming companies might monetize player-generated data in unexpected ways. Other location-based games and AR applications may follow similar models, using gameplay mechanics to collect valuable real-world datasets for commercial applications.
This development raises important questions about data ownership, consent, and transparency in the gaming industry. While technically legal under current terms of service agreements, the use of player data for training commercial AI systems may prompt discussions about clearer disclosure requirements and opt-out mechanisms for future data collection initiatives.