24/7 Service, Same Day Diagnostics

Our Latest Tweets

Facebook Is Developing Self-Learning Robots

Facebook researchers are creating robots that teach themselves to walk and satisfy their curiosity.


What used to be sci-fi fantasy is becoming reality as robots are programmed to react the same way as humans. Artificial Intelligence (AI) researchers who work for the social media giant, Facebook, developed an early prototype of robots that can teach themselves through trial and error and raw data. These revolutionary devices are much like human babies, learning to walk and explore the world around them and adapt to their surroundings.

How Robots Learn

Researchers from FAIR (Facebook AI research) said that the systems they create draw from raw data instead of structured information that is created for one specific task. With raw data, the devices can adapt to new tasks and circumstances. The ultimate goal is for the robots to move and explore their surroundings and manipulate the objects and environment around them.

Usually, AI can only solve problems that it has been programmed to and will compare the information it gathers in a situation to previously stored data. In a world of automation, the bots can only comprehend what they see by comparing it to what they already know. In the instance of Facebook, the device is essentially on its own.

The bots follow a reinforcement-learning algorithm, which is a section of machine learning in which the machine is prompted to find the best possible behavior or path in order to receive a reward. Much like in the real world, there is no set answer as to how a problem can be solved. The bot must decide what to do next based on the task at hand.

The First Step Toward Innovation

The first trial that researchers put a robot through was the task of learning how to walk. The device was a six-legged robot and was designed with many sensors. The robot is required to figure out spatial awareness, location, and balance just by using these sensors. Essentially, using the reinforcement algorithm meant that the more experience the robot had attempting to use its legs and walk, the better it performed.

This innovation was built with the hope that the robot could learn to walk in a matter of hours rather than weeks. The ability to self-teach itself a task in a short amount of time could lead to high efficiency in the workplace. As a result, it could truly increase the learning curve for technology and humans alike.


Social media giant, Facebook, has a research team working on AI for future uses.

Curiouser and Curiouser

Facebook teamed up with New York University to create a robot that had curiosity built into its mechanisms. Instead of exploring the environment randomly, the robotic arm that was developed explored in a structured manner to satisfy its curiosity. This method of learning reduced uncertainty.

For example, a toddler may explore the house, but when they find a doorknob, they will turn it, push, and pull the door until the understand how it works and will no longer be curious about the workings of a door. The robotic arm is created to optimize its actions to ensure that there is no uncertainty in how something works.

Researchers claim that the robotic arm knows little about itself to begin with, and as it learns what parts of the arm need to be used, it can plan what to do next. The arm is rewarded for resolving uncertainty to encourage it to be curious, rather than create a set action that will be repeated. This could ultimately lead to a camera or device from Facebook scanning images for useful information.

Robots are Tactile Learners

Researchers decided to involve the sense of touch to the robotic arm’s repertoire. The arm was successfully able to touch a 20-sided die to understand what side it was touching. By touching, the arm could analyze an object for patterns through touch. The bot was able to manipulate small objects that it encountered completely unsupervised.

Facebook wants to create AI systems that learn for themselves in any scenario, digital or otherwise. With this capability, we may see robots walking among us grabbing objects off of the store shelves sooner rather than later.

Preparing for an Automated Future

Facebook is not the only company who is making headway in the race for intelligent robots. Many researchers are looking to create machines that can think, act, and reason as people do. While technology is evolving before our eyes, there is no substitute for a human professional, especially when it comes to data recovery. Our engineers at Secure Data Recovery have decades of experience working with all types of devices and operating systems. To start your data recovery case, call 1-800-388-1266 today.

Request Help
Call for Immediate Assistance
24 Hour Service Expert Hotline
Alternatively, you can also fill out
a request help form online
Submit Help Request
Article Search
Secured & Certified


We are