Pokémon go data powers delivery robots: niantic leverages ar insights
Niantic, the company behind Pokémon Go, is repurposing the wealth of augmented reality data collected from players to train delivery robots. The move, spearheaded by Niantic Spatial, aims to solve a persistent problem for autonomous vehicles: reliable navigation in complex urban environments.
From pokéstops to pathways: ar data fuels robot navigation
The company has partnered with startup Coco, equipping its fleet of 1,000 suitcase-sized delivery robots with the spatial understanding gleaned from billions of player scans. This isn't just about recreating the familiar world of Pokémon Go; it's about translating the complexities of player interaction with augmented reality into a practical solution for real-world robotics. Niantic Spatial founder and CEO John Hanke explains, “Getting Pikachu to realistically run around and getting Coco’s robot to safely and accurately move through the world is actually the same problem.”
The data originates from consciously recorded and uploaded video scans of specific locations within the game – PokéStops and Gyms, essentially – allowing the robots to learn from the visual information players have already provided. Niantic claims access to 30 billion images of urban environments, a massive dataset that provides a rich training ground for Coco’s robots operating in cities like Los Angeles, Chicago, and Helsinki. This is a far cry from relying solely on GPS, which often falters in dense urban canyons, leading to significant positional drift.
The challenge, as CTO Brian McClendon points out, is that GPS accuracy in urban areas can be severely compromised. “If you look at that blue dot on your phone, you’ll often see it drift 50 meters, which puts you on a different block going a different direction on the wrong side of the street.” This spatial uncertainty makes autonomous navigation difficult. The goal is to give these robots a similar level of spatial awareness to a player navigating the game.
The implications extend beyond delivery services. If robots are to integrate seamlessly into human environments, they need a comparable understanding of their surroundings. The data gathered from Pokémon Go players provides a foundational spatial map that could revolutionize how autonomous vehicles navigate crowded streets, ultimately enhancing safety and efficiency.
The focus isn't merely on avoiding collisions; it’s about enabling robots to recover from unexpected bumps and jostles, mirroring the way players instinctively reorient themselves in the game world. This move signals a significant shift – the audience for augmented reality technology may soon be robots, not people.
The success of this venture hinges on how effectively the vast dataset of player-generated data translates into robust robotic navigation. For now, it's a bold experiment with the potential to reshape urban logistics and redefine the relationship between humans and autonomous systems.