Boeing testing visual data technology that could help predict the future, and protect us

Defending America
This is an archived article and the information in the article may be outdated. Please look at the time stamp on the story to see when it was last updated.
Data pix.

HUNTSVILLE, Ala. - Boeing is the world's largest maker of airplanes, but in this competitive world the factory line for "Big Blue" can always get more efficient, and safer.

"Protecting the future is all about looking at the art of the possible. Trying to anticipate the needs of the company before they're even needs," said Steve Swaine, the Director of Research and Technology for Boeing in Alabama.

Swaine is talking about a strange-looking video screen where objects, people and chairs for instance, are outlined with green boxes with a caption with objects' identity.  The screen and computer that generates it are found at Boeing's Machine Perception Lab in Huntsville. The green boxes indicate the computer recognizes and tracks the things it sees.

"It's all about taking visual data into the computer and then using machine learning in order to understand those visual images and be able to tell stories about what we see through the computer learning," said Swaine.

Boeing wants to track everything  at its manufacturing facilities.  That includes everything from tools to people. The idea is to account for everything for efficiency's sake, and anticipate unsafe situations.  An example of that would be to spot a worker about to walk in front of a piece of moving equipment before the collision happens.  VISTA, or "Video to Story Translation to Analysis" will do that.

It could also be used to catch busy highways, keeping track of problem areas and even decide when to alter speed limits or change traffic signals.

"It comes down to getting the quantity of data that we need to be able to teach the computer about the types of objects that we're interested in," said Zach Jorgensen, a Machine Learning Technologist in the Machine Perception Lab.

The job of course is to teach the computer what it needs to know, and then it will be able to recognize the desired objects. In the intelligence field, the Boston Marathon bombing is an example of how VISTA might be used.

"What if we were able to take all of that imagery from cameras at banks, and individual cell phones, and be able to feed that into a computer and interpret that in real time... in minutes, not hours and days," said Steve Swaine.

VISTA would take that interpreted data and put it in story form.  That would help investigators find the guilty. VISTA, by the way, is coming soon. "Maybe not tomorrow, but in three years time we should be able to have a system ready for customers," said Principal Investigator, Jan Wan Pei.