1. Home
  2. >>
  3. Latest
  4. >>
  5. NVIDIA
  6. >>
  7. Nvidia researchers help robots work together with humans

Nvidia researchers help robots work together with humans



Nvidia researchers help robots work together with humans


Inside the largest industrial laboratories we can find robot and advanced machinery that is dedicated daily to the production of a certain type of goods, respecting the assembly line. In this window we also find the men, who nevertheless have different roles and are usually destined for the last stages of the process.

Researchers from Nvidia they therefore want to offer us a more ambitious vision, one in which robots collaborate with human beings, learning from the latter the ways in which they will be able to perform the required tasks. Science fiction? It could be a more realistic future than one might think, judging by Nvidia’s early results.

  NVIDIA Super: Revolution or Cold War?

Here’s how i robot they will learn from human beings

The latest project by Nvidia was born in the new robotics laboratories located in Seattle, which since their opening have focused on a twofold, ambitious question: is it possible to make machine and man work in proximity? Can robots learn from us?

The researchers derived the first possible answers to these questions, presenting the results to theICRA 2018 (International Conference on Robotics and Automation) in Brisbane, Australia. The engineers would in fact be able to “teach” robots to observe a series of basic actions performed by humans, in order to replicate those same actions.

  NVIDIA recalls the vast majority of Shield tablets

As stated by Dieter Fox, senior director of Nvidia’s robotics research division, to TechCrunch’s Frederic Lardinois, the team would already like to enable this next generation of robots to work safely alongside a human. The machines would in fact be able to detect the presence of people, track their activities and understand how they can help them.

Of course, some researchers have already managed to train machines to perform repeated actions and even learn from the mistakes made by the machine itself, but – as Fox argued – the decision-making space for training robots in these same ways is too large to achieve. greater efficiency.

  Nvidia GeForce Now celebrates #GFN next Thursday with multiplayer

Efficiency that can instead be guaranteed by the new system created by the gi team Stan Birchfield And Jonathan Tremblay, which would allow researchers to teach a robot to perform new tasks by simply observing a human.

For researchers of Nvidia there is still a lot of work to be done to improve this system, but the foundations to better implement this technology have already been obtained. Among the team’s next goals is to teach machines to perform a wider range of tasks and to learn the vocabulary needed to describe those same activities.