Get our free email newsletter

Virtual Reality Trains Robots to do Tedious Tasks

Embodied Intelligence via YouTube.

Researchers at a startup are using virtual reality to train robots to perform simple tasks. This new method of teaching robots does away with the need to write a single line of code, making the learning curve far simpler for programmers and robots alike.

Founded by Pieter Abbeel, a professor of electrical engineering and computer science at the University of California, Berkeley, Embodied Intelligence, Inc. believes that virtual reality is the key to training robots. Alongside several of his students, Abbeel has constructed a virtual reality headset specially suited for this task. It comes with motion controllers that are connected to a robot named Brett. Brett is being taught numerous simple activities without the aid of coding.

The robot mimics the movements from the virtual reality as you perform them. This allows him to learn the motion patterns, matching them on his own. After a short teaching period, Brett is capable of performing the assigned task without the aid of a human.

- Partner Content -

Pulse Amplifier Definitions and Terminology

This application note serves as a comprehensive resource, defining key terms like duty cycle, pulse rate, rise/fall time, and pulse width, as well as discussing pulse on/off ratio, RF delay, jitter, and stability.

The virtual reality headset lets you see through a camera implanted in Brett’s head, giving you his exact perspective. This makes performing the tasks more interactive, with less room for mistakes or confusion. It takes Brett about 30 minutes to learn a task and perform it on his own.

In an interview with VRScout, Abbell explained the thought process behind this unconventional method of educating robots.

“The beauty of VR is you are virtually putting yourself into the robot, and see through the robot’s eyes. You get to move the robot’s hands, and teach it to learn the new skill in a way that is very natural.”

Pieter Abbeel, professor of electrical engineering and computer science at University of California, Berkeley

Although Brett has made remarkable strides in a relatively short period of time, he still has a ways to go before others like him are rolled out for consumers. There are still glitches to work out, and Embodied Intelligence believes they can streamline the learning process so it is even faster and more effective. But once these issues are dealt with, these robots could learn how to do menial work at factories and other such locations.

With this goal in mind, Embodied Intelligence has successfully raised $7 million in seed funding. Soon we should see smarter, adaptable robots capable of learning in ways we previously could never imagine.

Related Articles

Digital Sponsors

Become a Sponsor

Discover new products, review technical whitepapers, read the latest compliance news, and check out trending engineering news.

Get our email updates

What's New

- From Our Sponsors -

Sign up for the In Compliance Email Newsletter

Discover new products, review technical whitepapers, read the latest compliance news, and trending engineering news.