General Motors is developing next-generation autonomous vehicle systems that promise myriad benefits, from enhanced safety to greater efficiency. However, one of the biggest roadblocks in autonomous vehicle development is the software needed to quickly and accurately identify the environment around the vehicle while safely navigating through it. Now, it looks like GM is seeking a solution through a new license for artificial intelligence software system from the Department of Energy’s Oak Ride National Laboratory (ORNL).
The AI system is called the Multinode Evolutionary Neural Networks for Deep Learning, or MENNDL for short. According to ORNL, the software system “uses evolution to design optimal convolutional neural networks – algorithms used by computers to recognize patterns in datasets of text, images or sounds.” Although it can take expert coders up to a year or more to design effective neural networks, MENNDL AI can do the same job in a matter of hours, providing a system that allows autonomous vehicles to effectively analyze incoming data (such as onboard camera feeds), enabling more efficient energy use and increasing computing capacity.
“MENNDL leverages compute power to explore all the different design parameters that are available to you, fully automated, and then comes back and says, ‘Here’s a list of all the network designs that I tried. Here are the results – the good ones, the bad ones,'” said the head of ORNL’s Learning Systems Group and leader of the MENNDL development team, Robert Patton. “And now, in a matter of hours instead of months or years, you have a full set of network designs for a particular application.”
General Motors is the first company to obtain a commercial license for the award-winning MENNDL AI system.
In January of 2020, General Motors debuted its first fully autonomous vehicle, Cruise Origin, designed as a driverless robo-taxi with fully electric propulsion and twin-bench seating for next-generation ride-sharing transportation solutions.
Source: Oak Ridge National Laboratory