Further, Faster, Smarter: NVIDIA AI Powers Growing Student Racing Competition

More university teams rely on GPUs to cross the finish line at Formula Student Driverless race.
by Katie Burke

From rear wings to engine placement, certain design aspects of race cars are fundamental for speed. In the world of driverless racing, that element is NVIDIA.

At the third annual Formula Student Driverless racing competition this month in Hockenheimring, Germany, 70% of the teams — including the winning team ETH Zurich — relied on NVIDIA GPUs.

The high-performance, energy-efficient AI hardware made it possible for the university teams to run a wide array of algorithms in-vehicle, navigating the track at racing speeds.

The FSD competition is a growing part of the FSG international design event, which draws 4,000 students from 25 countries.

FSD is the biggest event of its kind in central Europe, garnering support from automakers such as BMW, Daimler and Volkswagen as well as suppliers such as Bosch, Continental and ZF.

Building Speed

Crossing the finish line is just half the challenge. For many teams, success starts in the garage.

Before they even begin building the car, the student teams create a concept and business plan for their vehicles. From there, they build race cars in a matter of weeks, ones that can withstand intense technical inspection as well as “scrutineering” — rigorous oversight from the competition’s officials.

To be cleared for the event, the vehicles must pass emergency brake and remote emergency system tests — meaning a human outside the car can take over controls — as well as be able to drive an unknown track and complete 10 laps on a set course.

The rigorous vetting process slimmed the field down to 20 university teams. Most used NVIDIA for their designs.

A Winning Formula

Racing a car without a human driver requires high performance computing that can run a diverse array of algorithms to navigate a course, as well as redundancies to ensure there are no failures. The platform must also be energy efficient, taking up as little space and power in the vehicle as possible.

To solve this challenge students turned to NVIDIA, with winning results. Fourteen of the 20 finishers used NVIDIA to run their self-driving system, with ETH Zurich bringing home the gold.

Courtesy of FSG

The benefits of high-performance compute were recognized across borders. German university Ravensburg CSU teamed up with Oregon State University to compete in the FSD event. With NVIDIA under the hood, the international collaboration was able to snag a place in the top 10.

From its origins as a niche race with just four teams a few years ago, FSD has grown into a major international racing event. And as more players enter the field with NVIDIA-powered vehicles, the competition continues to go further and get smarter, faster.