Caterpillar Developed a Big Data Infrastructure Using Deep Learning and Machine Learning - MATLAB
Video Player is loading.
Current Time 0:00
Duration 1:21
Loaded: 2.75%
Stream Type LIVE
Remaining Time 1:21
 
1x
  • Chapters
  • descriptions off, selected
  • captions off, selected
      Video length is 1:21

      Caterpillar Developed a Big Data Infrastructure Using Deep Learning and Machine Learning

      Larry Mianzo, Caterpillar

      Caterpillar has developed a big data infrastructure that provides a web-based ground-truth interface, a database for storing and querying ground-truth metadata, and an engineering interface with tight integration with MATLAB® products for machine learning, visualization, and code generation.

      Published: 22 Dec 2020

      [MUSIC PLAYING]

      I know there's a lot of excitement in the automotive industry for autonomous vehicles. But Caterpillar has been working on autonomous haul trucks for years. We also are working on things like non-line of sight semi-autonomous operations.

      So a lot of our environments are very harsh in some places we even have trouble getting operators to go into those environments. So we're getting the people out of those dangerous and harsh environments and allowing them to work at a distance in a much more comfortable setting.

      So in this case, you'd have an operator that could be miles away from the site. They could be gently supervising multiple vehicles, in this case some bulldozers.

      We do automatic ground truthing, so we can reduce the need for a human to go in and tediously label data. So we go from having to hand-label 100% of our data down to only labeling 80% or 90%. And the beauty of it is, as the classifiers get better and better, we have to label less and less data. So it's really been a tremendous advantage in efficiency.

      [MUSIC PLAYING]

      Related Products

      View more related videos