Facebook Discloses What Is Cooking In Its Secret Robotic Division

Facebook has been showing interest in robotics from a long time and now it has finally revealed the projects that are in progress.

Three projects that the company is working on are expected to solve the major challenge of artificial intelligence that needs a huge amount of labeled data for training. To address the issue, Facebook is carrying out research regarding teaching robots like humans around the world.

Robert Calandra, a research scientist at Facebook AI division said that they are creating algorithms that will be more reliable, robust and capable of learning fast.

An independent section, artificial intelligence research division, within the company is allotted for these projects, whereas research scientist does their work. Instead of focusing on the user-facing products, the company claims that they are carrying out robotic and artificial intelligence research to upgrade the industrial and academic level AI.

One of the projects is related to developing an algorithm through which six-legged robot will be able to learn how to walk in much less time, without feeding information about the environment.

Calandra said that the purpose of it is to make robots learn to walk within a few hours and achieve desired results without putting in data for training.


Curiosity in human beings let them learn more quickly and the same idea will be used for robots. The curiosity-driven technique is used by the company for applications of robotics arms and simulations.

The third idea focused by the company is to make robots learn by touching. A method has been created in which no training data is used and robots are made to perform tasks by learning through touch. Building accurate touch-enabled sensors hardware is the difficult part due to which machines often face challenges.

Many of the AI systems require labeled data for training which is not only time-consuming but may not be available at times. Driven by this, Facebook is working to train machines as humans are trained.
"Robotics research projects ... will help us build AI that can learn more efficiently and better generalize to new applications, even in noisy and highly complex environments such as the physical world. We are focused on using robotics work that will not only lead to more capable robots but will also push the limits of AI over the years and decades to come. If we want to move closer to machines that can think, plan, and reason the way people do, then we need to build AI systems that can learn for themselves in a multitude of scenarios — beyond the digital world.", explained Facebook AI team in a blog post.
Facebook is competing with companies like Google and Apple in the race of being the AI leader. Prominent firms of Silicon Valley including Facebook are spending enormously in AI. A 70 percent increase from 2017 has been noticed as the business value generated from artificial intelligence reached $1.2 trillion in 2018, as assumed by a market research firm, Gartner.

Advancing AI by teaching robots to learn
Photo: Facebook

Read next: “I am afraid of humans, not Artificial Intelligence”, says Google's Chief Decision Scientisthttps://www.digitalinformationworld.com/2019/05/should-we-fear-ai-here-is-what-cassie-kozyrkov-google-data-scientist-says.html

No comments:

Post a Comment