A massive dataset, comprising 30 billion images and accumulated over a decade by Pokemon Go players, is now being leveraged to train an advanced artificial intelligence navigation system. This unexpected development highlights how crowd-sourced gaming data can have significant real-world applications in robotics and technology.
How Niantic Spatial Developed the Visual Positioning System
Niantic Spatial, a spin-off company from Pokemon Go creator Niantic, has utilized this extensive dataset to create a Visual Positioning System (VPS). The VPS offers highly accurate location pinpointing within centimetres, making it particularly useful in environments where traditional GPS signals are weak or blocked, such as indoors or in dense urban areas.
Partnership with Coco Robotics for Enhanced Navigation
US-based firm Coco Robotics has partnered with Niantic Spatial to integrate this VPS technology into its delivery robots. This collaboration aims to improve navigation in complex city environments, allowing robots to maneuver more efficiently and reliably through streets and buildings.
Player Contributions Through In-Game Features
In-game features like 'Field Research' encouraged Pokemon Go players to scan real-world landmarks, such as parks, buildings, and other points of interest. These scans contributed to the detailed 3D models that form the backbone of the VPS, enabling the AI to recognize and navigate physical spaces with unprecedented precision.
This case demonstrates how recreational activities can inadvertently support technological advancements, with millions of players globally contributing to a dataset that now powers cutting-edge robotics. The implications extend beyond delivery services, potentially influencing future developments in autonomous vehicles, augmented reality, and smart city infrastructure.
