What started in 2020 as a simple task to earn some extra Poké Balls — pointing your cell phone at a monument and doing an augmented reality “scan” — became the basis of what is now considered the most advanced geospatial artificial intelligence infrastructure on the planet. This phenomenon of “invisible micro-work” has allowed a video game company to accumulate a treasure trove of data that Silicon Valley giants have been trying to replicate for decades without the same success, and they have achieved this with the collaboration of millions of people without knowing it.
Last week, Niantic Spatial (the new face of the creator of the Pokémon Go phenomenon) confirmed that its vast database of 30 billion images geolocated data is being used to guide autonomous delivery robots in real cities.
The announcement marks one of the largest value transfers in the history of technology: transforming the leisure and physical effort of millions of players into a commercial asset of incalculable strategic value for the robotics and logistics industry.
From the game to the road, which allows you to surpass traditional GPS
The announcement reveals the strategic partnership between Niantic and Coco Robotics. This company’s small robots, which already operate in dense cities like Los Angeles and Helsinki, faced a classic autonomy problem: so-called “urban drift.” Traditional GPS, dependent on satellites, has a margin of error that can reach five meters and fails drastically near tall buildings that block the signal.
Now, instead of relying on distant satellites, robots now use the Large Geospatial Model (LGM) from Niantic. This AI model works as a “spatial brain” that “sees” the world through visual memory created from what was captured by players.
Because every park bench, statue or storefront has been photographed and mapped thousands of times, from different angles, shadows and weather conditions, AI creates a “digital twin” of the world. By comparing what it sees in real time with this ultra-detailed map, the robot is able to determine its position with an accuracy of just a few centimeters, allowing it to safely navigate between pedestrians and obstacles on narrow sidewalks.
The triumph of “gamification” – and free labor
The scale of this AI training is unprecedented and redefines the concept of mapping. While companies like Google or Apple rely on fleets of camera-equipped cars LIDAR At a very high cost, Niantic achieved something that not even the biggest budgets can buy: the “last mile” (or rather, the “last centimeter”) of mapping done by human beings. While mapping cars are limited to roads, Pokémon Go players have entered parks, pedestrian plazas and alleys where vehicles do not have access.
It is estimated that, collectively, players traveled hundreds of millions of kilometers performing these scans for free. And more than the quantity, it’s the frequency that’s impressive: a Street View car passes through a street once a year or so, but a popular PokéStop is mapped several times a day. This constant update allows Niantic’s AI to understand structural changes in the city in near real-time.
However, the ethical debate becomes inevitable: will the player who dedicated his time to doing the scan of a monument in 2021 were you aware that you were, in effect, building the commercial navigation system of a logistics company in 2026?
Consent… And changing the rules
Niantic vigorously defends itself, stating that adherence to its services, which are themselves free, has always been voluntary (opt-in) and that the images are strictly anonymized, with faces and license plates blurred before entering the training model. However, for many data ethicists, there is a serious flaw in the transparency of long-term intentions.
The paradigm shift — going from “collecting data to improve the stability of virtual monsters” to “commercializing browsing data to third-party bots” — represents a gray area in terms of service. And these few users read them.
To make matters worse, from an ethical point of view, Niantic Spatial, after the strategic split that left the gaming division under the responsibility of Scopely, now has a monopoly on this visual map of the world.
One thing is certain: what was sold as a tool to make the game more immersive ultimately turned out to be the biggest robotic infrastructure and commercial surveillance project of the 21st century, built with the sweat of those who just wanted to “catch them all.”
Leave a Reply