CARS.COM — Leave it to us to find the slacker of the self-driving car world here at the 2017 Consumer Electronics Show in Las Vegas.
Positioned on a temporary test track adjacent to the sprawling Las Vegas Convention Center, the Audi’s task seemed simple enough. The short course consisted of two long corners at either end, an S-shaped curve in-between and a little loose gravel thrown in for good measure. What could go wrong?
Plenty, if it wasn’t for the “deep learning” capability that helps this Audi adapt to new surroundings. Developed in partnership with the tech firm Nvidia, this Audi Q7 project is meant to show how far a self-driving vehicle can go when relying on artificial intelligence.
There’s no elaborate sensor array on the roof or along the sides of this Q7, or extra cameras covering every available inch of sheet metal. This Audi is nearly identical to current production models, except for the fact that it relies on Nvidia’s Drive PX2 deep-learning platform.
The pared-back nature of the autonomous Audi shows how far this tech can go on its own, and how AI could speed up the arrival of self-driving cars. On the track, the Q7 didn’t miss a beat, though some of the steering inputs seemed overly complicated for the turns involved.
After two complete circuits of the track, a large construction barrier was placed in front of the Audi’s path. Using only front-mounted cameras to read and scan the road, the Q7 registered the obstacle and easily steered around it.
Why should I care? It’s worth mentioning that only hours before our ride in the Q7, one of Nvidia’s other self-drive test vehicles, a Lincoln MKZ sedan, had to slam on its brakes to avoid going off the very same course.
Audi is going on record as stating that AI computing will dramatically slash the time needed to bring self-drive vehicles to market. Audi’s Malhotra said the automaker will have “a highly-automated car by 2020.”
After a long day walking the massive convention halls at CES, it was a treat to let a car do all the driving while we relaxed in the backseat. But Audi’s three-year time frame remains extremely ambitious from an engineering standpoint in addition to the need to navigate legal hurdles that still face autonomous vehicles.