# Facebook Trains An AI To Navigate Without Needing a Map

**URL:** <https://onehack.st/t/facebook-trains-an-ai-to-navigate-without-needing-a-map/66773>\
**Category:** News & Articles\
**Tags:** social-media, ai\
**Created:** [January 24, 2020, 5:06am UTC](https://onehack.st/t/facebook-trains-an-ai-to-navigate-without-needing-a-map/66773 "2020-01-24T05:06:28Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![TheJoker](https://onehack.st/user_avatar/onehack.st/thejoker/32/22660_2.png) [@TheJoker](https://onehack.st/u/TheJoker)\
**Post date:** [January 24, 2020, 5:06am UTC](https://onehack.st/t/facebook-trains-an-ai-to-navigate-without-needing-a-map/66773/1 "2020-01-24T05:06:28Z")

</div>

A team at Facebook AI has [created](https://ai.facebook.com/blog/near-perfect-point-goal-navigation-from-25-billion-frames-of-experience/) a reinforcement learning algorithm that [lets a robot find its way in an unfamiliar environment without using a map](https://www.technologyreview.com/f/615078/facebook-has-trained-an-ai-to-navigate-without-needing-a-map/). MIT Technology Review reports:

 ![image](https://onehack.st/uploads/default/original/3X/a/0/a087dc133b2279f11c960bbc8fdf56f4ead0bc8e.jpeg)

- Using just a depth-sensing camera, GPS, and compass data, the algorithm gets a robot to its goal 99.9% of the time along a route that is very close to the shortest possible path, which means no wrong turns, no backtracking, and no exploration.

_This is a big improvement over previous best efforts. […] Facebook trained bots for three days inside [AI Habitat](https://aihabitat.org/), a photorealistic virtual mock-up of the interior of a building, with rooms and corridors and furniture. In that time they took 2.5 billion steps – the equivalent of 80 years of human experience._

Others have taken a month or more to train bots in a similar task, but Facebook massively sped things up by [culling the slowest bots from the pool](https://techcrunch.com/2020/01/21/facebook-speeds-up-ai-training-by-culling-the-weak/) so that faster ones did not have to wait at the finish line each round.

> As ever, the team doesn’t know exactly how the AI learned to navigate, but a best guess is that it picked up on patterns in the interior structure of the human-designed environments. Facebook is now testing its algorithm in real physical spaces using a LoCoBot robot.