Self Driving Challenge 2023: won by RUG, learned by all

Sounds logical, but the technology used seems overwhelming. The six participating teams use the same electric kart and make up the difference with self-written software. The vehicle may not look as flashy as a Tesla, but the technology is more interesting than a first glance might suggest. "The kart, like a normal car, has a CAN network," explains Ian Matroos, Data Scientist at the RDW. "It works exactly the same, again safety systems take priority over relative side issues." The kart is equipped with three Logitech front cameras and a rotating LiDAR sensor from Slamtec. Computing power comes from an Intel NUC, a compact but common PC.

Last year, when Zwolle-based Windesheim University of Applied Sciences took the top prize, the kart had only one camera. The two new, side-facing ones are gratefully used by the teams. That does not apply to the LiDAR. "That is therefore mainly intended for obstacle detection," explains RDW software engineer Thijmen Fellinger. "But because driving a complete lap is already challenging enough, we chose not to use obstacles yet this year." So much for engineering, it's time to drive!

Off! After the starting signal from race director Joël Coster, the kart from Groningen University of Applied Sciences takes off. The very first drive of the Self Driving Challenge starts at a walking pace, around 4 km/h. The kart moves confidently along the track and rounds the first curve without any problems, which it failed to do last year. In the second curve, things go wrong: the left front wheel crosses the line. Team captain Martijn: "The sun has a lot of influence, we are set for cloudy conditions. Yesterday, during qualifying, we made it to the penultimate curve." During the final, it is not reached, though the final drive - with adjusted settings - is good for a podium place.

For The Hague University of Applied Sciences, the first corner still proved challenging, their kart diving to the right at right angles well before the turn. After several attempts, during which the kart stranded at the same point, another software model was loaded. This gets the vehicle slightly further, but not far enough. The Hague University of Applied Sciences will finish in sixth and last place.

Great is the contrast with the University of Groningen. Although this team's vehicle gets off to a somewhat hesitant start, it drives on as if autonomous driving is the most natural thing in the world. For a moment it is tense when the kart stalls, but it quickly recovers. After reaching the finish line, the team even puts in a flawless second lap. Their relatively simple line detection trumps complicated neural networks and machine learning solutions from the competition.

Despite the Groningen university team's achievement, the final result is far from certain. During qualifying, the University of Twente team also managed to cover the entire circuit autonomously. During the race, things did not go so well; for a while, the question even arose whether the kart would function at all. After a complete reset of the software, it starts anyway. The drive starts hopefully, but ends in the grass of the chicane. End of practice. The emotion among the team is tangible, members fall into each other's arms comfortingly. Supervisor Luuk Spreeuwers: "They have focused on this enormously, it is extra disappointing that two good laps were driven yesterday. We will fortunately be allowed to do another round later, outside the competition."

Now it is the turn of Rotterdam University of Applied Sciences. Their vehicle struggles to get going and experiences several strandings in the first corner. With several software adjustments, they manage to reach the fourth corner, where the kart opts for an off-track route. There the final ends for Rotterdam, good for fourth place.

Finally, it is the turn of Windesheim University of Applied Sciences, last year's winner. Although the defending champions did a lot more programming than their predecessors, the podium is not in the cards this year. They reach the third bend and thus have to settle for place five.

After the last final drive, the Groningen university team demonstrates what an advantage they really have by once again driving around the circuit at higher speed. Afterwards, the students from Twente also retaliate by making a full lap all by themselves. This way, they return home with a slightly better feeling.
 

With all the racing action, you would almost forget that an 'autonomous festival' has been rigged up next to the track. Qbuzz is there showing a green, articulated bus from the Columbuss project for autonomous urban and regional transport. With a semi-autonomous Tesla Model 3 and Mercedes-Benz EQE, demonstrations will be given on public roads. NXP's Volkswagen ID.4 also stands out because of an extra frame with radar sensors developed in-house. "A radar, unlike a camera or LiDAR, is insensitive to weather conditions," explains head of communications Martijn van der Linden for choosing precisely this technology. To casually add: "Projects like the Self Driving Challenge are hugely important for technology in the Netherlands, actually a minister should be on stage here."

Where is the Self Driving Challenge headed? If it's up to Gerard Doll, Division Manager Vehicle, Regulation & Admission at the RDW, there will definitely be obstacles on the track in the future. "I also think the technology could be more complex, with the addition of a radar sensor, for example. In addition, we are thinking of an open category, so that other autonomous vehicles can participate." Why the RDW is organising this challenge? "This is the way for us to get to know a new generation of engineering students, but also the perfect opportunity for students to learn about the world of self-driving vehicles." That it works is proven by Natasja, student from team The Hague University of Applied Sciences: "This is our first contact with this technology and I can't really think of anything else now!"