Sign In

Communications of the ACM

ACM TechNews

How to Train a Robot (Using AI and Supercomputers)


View as: Print Mobile App Share: Send by email Share on reddit Share on StumbleUpon Share on Hacker News Share on Tweeter Share on Facebook
Examples of three-dimensional point clouds synthesized by a progressive conditional generative adversarial network for an assortment of object classes.

Computer scientists at the University of Texas at Arlington developed a deep learning method to create realistic objects for virtual environments that can be used to train robots.

Credit: William Beksi, Mohammad Samiul Arshad/UT Arlington

Computer scientists at the University of Texas at Arlington (UT Arlington) are using generative adversarial networks (GANs) to train robots about objects.

Such training typically requires a large dataset of images, but GANs can create a potentially limitless amount of data with which to train a robot in just seconds.

The researchers developed PCGAN, the first conditional GAN to generate dense colored point clouds in an unsupervised mode.

In an evaluation of 5,000 random samples for each object class, the researchers determined PCGAN can synthesize high-quality point clouds for a disparate array of object classes.

Said UT Arlington's William Beksi, "We're starting small, working with objects, and building to a hierarchy to do full synthetic scene generation that would be extremely useful for robotics."

From Texas Advanced Computing Center
View Full Article

 

Abstracts Copyright © 2021 SmithBucklin, Washington, DC, USA


 

No entries found