Final week, Niantic announced plans to create an AI mannequin for navigating the bodily world utilizing scans collected from gamers of its cell video games, equivalent to Pokémon Go, and from customers of its Scaniverse app, stories 404 Media.
All AI fashions require coaching knowledge. To this point, corporations have collected knowledge from web sites, YouTube movies, books, audio sources, and extra, however that is maybe the primary we have heard of AI coaching knowledge collected via a cell gaming app.
“Over the previous 5 years, Niantic has targeted on constructing our Visible Positioning System (VPS), which makes use of a single picture from a telephone to find out its place and orientation utilizing a 3D map constructed from folks scanning fascinating places in our video games and Scaniverse,” Niantic wrote in an organization weblog submit.
The corporate calls its creation a “massive geospatial mannequin” (LGM), drawing parallels to massive language fashions (LLMs) like the sort that energy ChatGPT. Whereas language fashions course of textual content, Niantic’s mannequin will course of bodily areas utilizing geolocated pictures collected via its apps.
The dimensions of Niantic’s knowledge assortment reveals the corporate’s sizable presence within the AR area. The mannequin attracts from over 10 million scanned places worldwide, with customers capturing roughly 1 million new scans weekly via Pokémon Go and Scaniverse. These scans come from a pedestrian perspective, capturing areas inaccessible to automobiles and street-view cameras.
First-person scans
The corporate stories it has educated greater than 50 million neural networks, every representing a selected location or viewing angle. These networks compress 1000’s of mapping pictures into digital representations of bodily areas. Collectively, they include over 150 trillion parameters—adjustable values that assist the networks acknowledge and perceive places. A number of networks can contribute to mapping a single location, and Niantic plans to mix its data into one complete mannequin that may perceive any location, even from unfamiliar angles.