Nvidia offers to train robots in the emulator, Isaac

At the conference, GTC 2017, Nvidia introduced the development in the field of robotics, which are said to dramatically simplify the creation and training of intelligent machines.

We are talking about the emulator, Nvidia Isaac, which means graphics technology creates a virtual world, designed for training of robots.

Testing and training using a real prototype environment can be expensive and impractical inventions and re-create the environment in which to operate the robot, sometimes it’s too dangerous and difficult. The simulation resolves all these issues.

Isaac serves as the basis for a modified engine Epic Games Unreal Engine 4. Using technologies for deep learning, it is possible to implement comprehensive test scenarios and training. According to Nvidia, the result of a few minutes manages to give the robot knowledge, which in the ordinary procedure would require several months. The results are subsequently transferred to the real system, intended for introduction.

In addition, were presented to the reference samples for all robots on the Jetson platform.

During GTC more than five dozen companies showed robots that use Nvidia Jetson. As you know, the Jetson TX1 module was presented in November 2015 and in March of this year saw the light module Nvidia Jetson TX2.

Source: Nvidia



(Visited 21 times, 1 visits today)
No tags for this post.