In business, innovation is essential for growth and staying ahead in a competitive market.
It drives new ideas, improves processes, and creates products that meet evolving customer needs. Without innovation, businesses risk stagnation and losing relevance. Without innovation, even the most innovative
Successful companies are those that know what consumers want, seize opportunities, adapt to the changes, and secure the position as one of the key players in their respective industry.
But smart companies are the ones that are able to maximize the potential of their existing resources and leverage them to their fullest.
Niantic Labs, the privately-held Pokémon GO app parent company, acquired the popular app to the masses in July of 2016, and had over a billion installs before 2019.
And here, the company has been quietly, and cleverly using data from Pokémon GO players to build a "Large Geospatial Model."
ICYMI - we’re building a Large Geospatial Model (LGM) @NianticLabs to achieve spatial intelligence which we believe is critical for ushering in the new era of spatial computing and the AR glasses that are right around the corner.
As part of our Niantic Visual… pic.twitter.com/IEEUMNUsF5— asim ᯅ (@tweetsfromasim) November 13, 2024
Pokémon GO requires players of the game to travel from one place to another in the real world, to find various Pokémons and battle other players.
Thanks to using augmented-reality (AR) technology, it was able to provide a unique experience, as players have to explore new locations that they probably have never ventured to, and socialize like they never did.
As time progressed, Niantic Labs managed to acquire a substantial amount of data from users' activities.
The company did this by strategically placing PokéStops and Pokémon Gyms, and with the many players visiting the venues, Niantic is able to get its hands on a more detailed and granular model of streets, sidewalks, parks, and businesses.
It was able to grab images and scans of locations that even Google doesn’t have street view images of.
This is possible because Pokémon GO relies on players walking around with their phones, rather than hiring people to walk or drive with specialized cameras.
The type of data Niantic gathers is referred to as the Visual Positioning System (VPS).
Unlike GPS (Global Positioning System), which uses satellite signals to determine a device's location, VPS relies on visual data, such as images and scans of the surrounding environment. By analyzing landmarks, building shapes, and other physical features, VPS can provide highly precise location information, in ways that GPS cannot.
In all, VPS data is built using a single image from a phone to determine its position and orientation using a 3D map built from Pokémon GO players scanning the areas around them.
Long story short, Pokémon GO players have helped Niantic create an intelligent machine that maps the entire planet, all for free.
VPS data is particularly valuable for enhancing AR apps like Pokémon GO itself.
But at the same time, thanks to the data that include accurate placement of digital objects in the real world, Niantic is able to built a Large Geospatial Model, or LGM.
This name, the company explains, is a direct reference to Large Language Models (LLMs) Like OpenAI’s GPT, which are trained on vast quantities of text scraped from the internet in order to process and produce natural language.
Niantic explains in a dedicated web page, that a LGM aims to achieve the same thing as LLM, but for the physical world.
The company said that the technology "will enable computers not only to perceive and understand physical spaces, but also to interact with them in new ways, forming a critical component of AR glasses and fields beyond, including robotics, content creation and autonomous systems. As we move from phones to wearable technology linked to the real world, spatial intelligence will become the world’s future operating system."
By training an AI model on millions of geolocated images from Pokémon GO from players around the world, Niantic aims to create a system that can learn to predict and understand its immediate environment.
Just like how LLMs can generate coherent and convincing sentences by statistically determining the most likely sequence of words, LGM can do the same but with geospatial data.
From more than 10 million scanned locations around the world, and over 1 million of those are activated and available for use with its VPS service, Niantic said that it has trained more than 50 million neural networks, with more than 150 trillion parameters.