A blog post on Niantic's website has captured the attention of many Pokemon Go fans and tech savvy people, as the company details its efforts to train machine learning-powered geospatial models through its app. That means you, yes you, probably had a role in setting it up.
The blog post, titled “Building Large Geospatial Models for Spatial Intelligence,” notes that through Niantic's visual positioning system, it has collected a lot of data and do extensive training on their geospatial models (thanks to 404 Media).
iGamesNews
“As part of the Niantic Visual Positioning System (VPS), we have trained more than 50 million neural networks with more than 1.5 trillion parameters, capable of operating on more than 1 million locations. In our analysis of the Large Geospatial Model (LGM) )’s vision, each of these local networks will help form global large-scale models, enabling a shared understanding of geography and making sense of places that have not yet been fully scanned.”
Okay, there are probably a couple of big issues here. First, what does this geospatial model actually do? Simply put, it will help computers perceive and navigate real-world environments. What this will actually mean for technology, as mentioned in the aforementioned blog post, is that things like AR glasses, robots, and autonomous systems will be able to better navigate across streets without having to struggle to identify which lamppost they're heading toward.
New: the company behind Pokemon Go has announced it is using data collected by its millions of players to build an AI model that can navigate the real world, and could be used for robots. Doubt Pokemon Go players anticipated contributing data to this project www.404media.co/pokemon-go-p…
— Joseph Cox (@josephcox.bsky.social) 2024-11-19T15:36:06.741Z
Second question, what is this visual positioning system? Well, Pokemon Go players probably know this through the lens of the game's AR mapping tool. These have become part of field research missions and allow players to scan around the world for in-game rewards. As another Niantic blog post explains, VPS is a technology that can better map AR content to the real world. So, in simple terms, the game maps poke stations to local museums rather than pub toilets.
Niantic says that users have currently scanned more than 10 million locations across its many games, and even provided heat maps to show that most of the data was collected in Japan, coastal US and central Europe. Artificial intelligence and machine learning models get a well-deserved bad wrap for being used in a bunch of useless nonsense. But if you're involved in artificial intelligence development, this does show real technological progress as far as AR is concerned. One just hopes you're not using Pokemon Go to scan your entire house for Niantics, lest a real-life Metacross break down your bathroom door like they did in The Shining.
What do you think of all this? Personally, I am largely opposed to AI technology on an environmental level, at least until we find good solutions to the energy crisis, but maybe you feel differently! Let us know below.
