Niantic, the creators of the popular Pokemon Go game, have announced the development of a cutting-edge technology called the Large Geospatial Model (LGM), which aims to understand and predict characteristics of various areas by leveraging data from existing geographical sources.
Previously, Niantic offered the Visual Positioning System (VPS) which pinpointed phone coordinates at a centimeter level using single images. However, this process relied on pre-existing image databases, limiting the available data despite having a vast collection of images of different locations worldwide.
The LGM will harness knowledge gathered from various images collected by Niantic, such as numerous global landmarks. When the entrance of a particular landmark is recognized, the LGM can instantly predict how the rear of the landmark should appear, even if no images exist in the database. This approach enables the VPS service to predict phone coordinates from unseen angles.
It is anticipated that the LGM will have alternative applications beyond coordinate identification, such as creating three-dimensional models of areas or constructing game scenes through imagination.
Niantic has not disclosed a timeline for the completion of the LGM development or whether it will be available for public use. Previously, the VPS service was not open for general registration but required contact with Niantic representatives.
TLDR: Niantic is developing the Large Geospatial Model (LGM) to understand and predict features of areas using data from existing image sources, expanding beyond their previous Visual Positioning System (VPS) technology.
Leave a Comment