Pokemon Go players trained delivery robots

Pokemon Go players unwittingly became participants in a technological breakthrough: more than 30 billion images they took formed the basis of a navigation system for delivery robots. Niantic's development is already being used in partnership with Coco Robotics and is being tested on city streets.

How Pokemon Go helped train robots

Niantic announced that it has collected over 30 billion images thanks to the activity of Pokemon Go players, writes xrust. This data formed the basis of visual positioning technology (VPS).

Key features of the system:

• determines the location based on surrounding objects, rather than by satellite
• works in conditions where GPS loses accuracy
• achieves accuracy of up to several centimeters
• uses images of buildings, streets and landmarks

Players, without realizing it noticing, they created a huge database, scanning real objects for in-game bonuses.

Partnership with Coco Robotics

This week, Niantic announced a partnership with Coco Robotics, a maker of delivery robots.

New robots will:

• move along the sidewalks
• use VPS instead of traditional GPS
• navigate using 4 built-in cameras

class=»notranslate»>__GTAG10__ deliver food and goods over short distances

According to the company, this will significantly improve the accuracy of delivery in dense urban areas.

Why GPS no longer works

Traditional GPS systems often fail:

• in narrow streets and between high-rise buildings
• in tunnels and closed spaces
• in bad weather conditions

VPS solves this problem by “comparing” the real image with an already existing 3D map of the area.

Experts note that such technologies can become a new standard for urban navigation. According to industry estimates, VPS accuracy can reach 5-10 cm, which is tens of times better than conventional GPS.

How the “living map” of the world was created

Data collection has especially accelerated since the introduction of the object scanning feature in 2020.

Players could:

• scan statues and sights
• receive rewards for “field research”
• explore game “arenas”

Each this action helped create 3D models of the real world.

Thanks to the many users, the system took into account:

• different shooting angles
• lighting and time of day
• weather conditions

class=»notranslate»>__GTAG15__ height and distance

This made it possible to achieve high mapping accuracy.

What will happen next

Once robots hit the streets, the learning process will not stop. On the contrary:

• robots will start collecting new data
• cards will be constantly update
• navigation accuracy will continue to increase

Niantic calls this creation “living map” of the world — a dynamic system that is updated in real time.

Based on materials:

• https://nianticlabs.com
• https://www.theverge.com

Instead of an afterword

Pokemon Go released in 2016 and quickly became a global augmented reality phenomenon. Since then, Niantic has been actively developing AR and mapping technologies, turning game data into tools for real business and urban infrastructure.

Xrust Pokemon Go players trained delivery robots