Student setzt Miniaturauto auf Fahrbahn

The future of autonomy at 1:8 scale

Audi Autonomous Driving Cup 2018: a focus on artificial intelligence

What do carrots, rubber duckies, and the Village People all have to do with autonomous driving? They were all part of the Audi Autonomous Driving Cup 2018. Alongside some eclectic offerings, Audi blog author Andreas Wittke saw a wealth of both real and artificial intelligence on the carpeted driving course.

Miniaturautos Frontansicht
The model cars are ready for the Audi Autonomous Driving Cup final 2018. This year's focus was set on artificial intelligence.

The focus of this year’s Audi Autonomous Driving Cup (AADC) lay on the development of self-driving cars with artificial intelligence. The eight teams of students who made it into the final were required to demonstrate their skill by delivering a scientific lecture, then completing a handling course and an open challenge.

At the AADC, artificial intelligence means that the teams need to program their cars to “learn” by means of image recognition on the handling course, then use this information to steer themselves. The tricky part? One image is not sufficient; error-free execution requires a massive volume of images or data.

Artificial intelligence: data volume as a success factor

You could perhaps liken it to a child who is looking at a street crossing for the first time. It doesn’t just take in a single, still snapshot of the situation; it soaks up all visual — and sensory — impressions as well. That’s the only way it can create a complete picture of the situation and learn to understand it. So if the autonomous car only has a few images of a certain traffic situation saved, it can’t correctly analyze it, nor can it process and react to unexpected situations.

In the five-month preparation period, the students found creative and impressive solutions to these challenges — even if their self-driving Audi Q2 lookalikes occasionally drove across a line and caused a collision with other cars, objects, or Barbies.

Vogelperspektive auf Audi Forum

The Audi Forum is rather busy – some spectators have come to see the showdown of the model cars.

Automatic object recognition: not always intelligent

But let’s take a look at the competition. This year, a laser scanner was included in the Mini Audi Q2 for the first time. In addition, the teams were supplied in advance with high-definition maps of the handling course. An emergency vehicle was another new addition this year. The cars needed to recognize it, drive safely to the right-hand side of the road, and come to a stop, as well as giving it the right of way at crossings.

The greatest difference to last year’s course was a ramp whose incline posed a particular challenge for the car’s obstacle recognition programming. Overall, the teams brilliantly mastered the handling course on the lovingly designed carpeted track.

Networked autonomous cars: a brilliant solution for smart cities of the future

In the open challenge, the eight teams of students showed off their creative talents. In this competition, they presented a project of their own design. The only requirement was that it focus on artificial intelligence and networking the cars with each other or with smart devices.

The jury was most impressed by the AlpaKas’ concept: a car-to-car solution for unimpeded driving. The software notifies the other cars of, for example, an accident or other disruption in traffic flow so that they can intelligently update the route to their destination. “An example that could be used immediately in real life,” said juror Harald Altinger.

Other cases showed examples of autonomous taxis or virtual towing assistance. Two teams stood out with their humorous presentations. The Cariosity team from the University of Wedel quickly transformed their car into a virtual pet that, at the push of a button, got excited and wagged its tires.

Zuschauer schauen zu und klatschen
The guests did more than just watching – the student teams involved them in the track, the university team from Erlangen-Nürnberg let the spectators dance the YMCA.

They also attached a rubber ducky to the car that drove after a remove-controlled carrot. The FAUtonomous team from the University of Erlangen-Nuremberg got the audience involved; they had to complete simple tasks like doing the YMCA dance. The car continued driving only after it recognized that the task had been completed.

Globalization in miniature: an international AADC competition

This year, universities from Italy, England, and Austria participated for the first time, although none of them made it to the finals. “I’m happy that the Cup has established itself at the universities — and has even been incorporated into the lectures. This lets us build a relationship with young talent early on,” explains Florian Knabl, AADC sponsor.

3 Miniaturautos beim Parcours
Full steam ahead: the cars during the course.

The Cariosity team from the University of Wedel, led by Felix Maass, was also enthusiastic: “It’s our first time here, and we definitely had some initial difficulties. But the Cup is a cool event and we have all learned an incredible amount from one another. The atmosphere and relationship between the teams is relaxed and friendly.”  Shih-Jan Lin already participated last year, winning the competition as part of the OptILmDriveX team from the Technical Univeristy of Ilmenau. “It’s great to be part of this again. The challenges this year were even greater than before; the new handling course, in particular, presented us with difficulties.

Programmers wanted: jobs are in store for many participants

Thanks in part to their success in the open challenge event, the AlpaKas from Karlsruhe ultimately prevailed. 10,000 euro in prize money made their victory even sweeter. Knabl explains the true value of the competition: “The AADC is a win-win situation for both sides. The students not only have the chance to win something; they also learn to manage an entire project and get a glimpse into the real-life working world. And we profit from the fresh ideas and potential job applicants.”

Zuschauer

The little ones were excited about the model cars as well.

Around 15 participants from the previous four years are now working at Audi — as student employees, trainees, and even in permanent positions. “We were able to hire many of the developers immediately after the competition. They worked in our real-life development environment for six months and really know their way around now,” explains juror and co-founder of the AADC, Lars Mesow.

High-quality programming: achieving success through savvy

This year’s competition was already the fourth Audi Autonomous Driving Cup. Mesow is excited about the way the competition has developed: “The level of sophistication and the quality of the programming for all teams in the finals this year was very high. I always find it fascinating to see the different perspectives on autonomous driving that the students bring to the table.” With a grin, he adds: “We can learn a lot in this regard, because we sometimes see the world through our Audi lenses, and that leaves us blind to certain topics.”

A summary of the Audi Autonomous Driving Cup 2018

Ten university teams, numerous model cars and ambitious participants – the Audi Autonomous Driving Cup kicked off with an event in June during that participants received basis software and two model cars. After a testing event in October, the finale took place in Ingolstadt in November. On the finals day, eight teams were competing for the win – in the end, the team AlpaKa from the Karlsruhe Research Center for Information Technology secured the win and took the prize money home to Karlsruhe. The Audi Blog takes a look at the competition:

Barbie crossing: the model cars were programmed to avoid any obstacles.
The jury during their evaluation.
In the end, the AlpaKa team convinced the jury. The student team presented a car-to-car for unimpeded driving. The software lets other cars know about disruptions and guides them to their destination on alternative routes.

Barbie crossing: the model cars were programmed to avoid any obstacles.

1/3

The jury during their evaluation.

2/3

In the end, the AlpaKa team convinced the jury. The student team presented a car-to-car for unimpeded driving. The software lets other cars know about disruptions and guides them to their destination on alternative routes.

3/3
Audi Autonomous Driving Cup 2018
Audi MediaCenter

Team AlpaKa from the Karlsruhe Research Center for Information Technology clinch the title

The final of the Audi Autonomous Driving Cup 2018 saw Team AlpaKa fend off the challenge from seven other university teams to win the title. The prize money of 10,000 euros therefore goes to Karlsruhe.

Read more

Make a comment

Your email adress will not be published

1.9.0