LISTEN LIVE

Perfecting Driverless Cars Is More Ones And Zeroes Than Nuts And Bolts

Jan 2, 2020

We've heard it for years: driverless cars are the way of the future. But there’s lots of work to be done before self-driving cars rule the road.

Students and faculty at Florida Polytechnic University are trying to solve one of the trickiest problems: how to make a machine driver think like a human, but better.

When you drive, you make thousands of decisions you probably don’t think about -- like when to brake to avoid hitting the car in front of you or slowing down during a thunderstorm.

A mock-up of the ISEAUTO vehicle built by Tallinn Technical University.
Credit Bradley George/WUSF

“You kind of have to turn into a sort of language that the machine can understand,” said Chris Medrano, a graduate student in computer science at Florida Poly.

Instead of eyes, self-driving cars rely on sensors, like motion detectors. Medrano and his colleagues in the university’s Advanced Mobility Institute are testing how those sensors react to rain, pedestrians, cyclists, and billions of other scenarios.

“Not only do you have to know that it makes the right decision, but that it makes the right decision for the right reasons,” Medrano said. “So it's not just like it happens to randomly be able to pass the test."

You won’t see Medrano testing these sensors on I-4. He’s working in a computer lab with a mockup of an autonomous vehicle built by Tallinn Technical University in Estonia.

Assistant professor Suleiman Alsweiss said it’s important to run as many simulations as possible before trying out the technology in the real world.

“The amount of processing that we do while driving is just amazing, and hopefully one day we will transfer that knowledge to sensors, and they can do it even more efficiently,” he said.

All modes of transportation could benefit from Florida Polytechnic’s research. Planes, ships, and trains could one day use the same sensor technology as driverless cars.