© 2021 Innovation Trail

Cars may soon drive themselves

David Saddler
via Flickr

Many of us make calls, sip coffee, catch ourselves dozing off, or sneak a peek at the newspaper while driving.  But what if technology allowed us to pursue these activities without risking life and limb? Scientists at Yale University and New York University are working on a supercomputer that could eventually lead to driverless cars. The device will have the ability to navigate through traffic, and avoid objects in the road.

The supercomputer is designed to mimic the human visual system. The computer, called NeuFlow, is embedded in a single chip and uses complex mathematical language developed at New York University. Dr. Eugenio Culurciello, associate professor at Yale’s School of Engineering and Applied Sciences worked with NYU for nearly two years to develop the system. He also collaborated with other scientists that study the brain and applied the findings to computer technology. The result is an efficient chip that’s capable of running billions of operations per second.

“It’s inspired architecture. It’s a neural network that has multiple layers. It’s inspired by the way our visual system is in the brain. We have not really used it to navigate a car. We are in the early stages to prove that this can be used,” Dr. Culurciello said.

You could see driverless cars in the next five years according to Dr. Culurciello. The chip is being tested in video games to do specific tasks like spotting objects and pedestrians on the road.

Dr. Culurciello says the system could also be used in assisted living environments to monitor movements and call for help when an elderly person falls. Other applications include monitoring large crowds for security purposes.