Pokémon go data powers delivery robots: niantic leverages ar insights
Niantic, the company behind Pokémon Go, is repurposing the vast trove of augmented reality data collected from players to train delivery robots. The move, spearheaded by Niantic Spatial, aims to solve a persistent problem in robotics: reliable navigation in environments plagued by GPS inaccuracies.
From pokéstops to pathways: how ar data is teaching robots to navigate
The company has partnered with robotic startup Coco, equipping its fleet of 1,000 suitcase-sized delivery vehicles with the spatial understanding gleaned from billions of user-submitted video scans. This isn't some clandestine data grab; Niantic has openly discussed the need for real-world data to enhance its augmented reality maps, a necessary step toward a future saturated with AR Technology.
“Everybody thought that AR was the future, that AR glasses were coming,” said Brian McClendon, CTO at Niantic Spatial, in an interview with MIT Technology Review. “And then robots became the audience.” The problem, he explains, is the urban canyon – the dense concentration of buildings that severely disrupts GPS signals, causing delivery robots to frequently drift off course.
Niantic Spatial founder and CEO John Hanke elaborated, noting the surprising parallel between training Pikachu to navigate the real world and getting Coco's robots to do the same. “If robots are ever going to assimilate into that environment in a way that’s not disruptive for human beings, they’re going to have to have a similar level of spatial understanding,” he stated. The company possesses access to 30 billion images of urban environments, a dataset Coco’s robots are now learning from across cities like Los Angeles, Chicago, and Helsinki.
The data consists of consciously recorded video scans of specific real-world locations – a PokéStop or Gym, essentially – already identified within the Pokémon Go game. This allows Coco's robots to learn the visual characteristics of these locations and navigate accordingly, even when GPS is unreliable. The potential implications extend far beyond delivery services; a robust spatial understanding is vital for any autonomous vehicle operating in complex urban settings.
The Technology addresses a fundamental challenge in robotics: ensuring robots can accurately determine their position even when faced with obstructed GPS signals. The solution isn't about replacing GPS entirely, but augmenting it with a rich understanding of the visual environment.
Niantic's approach highlights a fascinating shift: the practical applications of AR Technology may be emerging not in wearable devices, but in the very robots that will populate our streets. The robots' ability to recover from jostles and bumps, a skill honed by years of Pokémon Go player data, is a testament to the unexpected ways in which technological innovation unfolds.
The sheer volume of data Niantic has accumulated – millions of scans submitted by players over the years – represents a valuable resource for developers seeking to create more intelligent and adaptable robots. It's a prime example of how data collected for one purpose can unlock possibilities for another.
Ultimately, Niantic’s bet is that the visual language of a game like Pokémon Go can translate into a more reliable and intuitive way for robots to navigate our increasingly complex world.