"Thanks to the competition, we realized that at least 38 universities worldwide would appreciate the know-how in the field of control systems that is taught in Prague, specifically driving dynamics control and its role in autonomous vehicles. That is why we have started the preparation of an intensive international course on this topic from 2025, in two versions - an academic one for university students and teachers and an industrial one for company experts," says doc. Tomáš Haniš.
The aim of the Self-Driving Car Student Competition was to develop a driving system for an autonomous car that could drive the specified route as fast as possible without any external intervention. However, other judging criteria included driving precision and timely response to traffic signs; in fact, there were two traffic lights and a stop sign along the route. Thus, the students had to implement features such as data fusion of individual car sensors, design of car dynamics control and route guidance, trajectory planning, and image processing systems for road and traffic sign detection. The actual implementation of the system then took place on a 1:10 scale vehicle platform.
Five Master's students from the Cybernetics and Robotics study program went to Canada for the competition together with the team captain, PhD student Jan Švancar, who works at the Department of Control Engineering at FEL CTU. The students divided the roles in the team according to their experience and hobbies. Jan Kohout, together with Jakub Macar, was responsible for the driving dynamics of the car and track guidance, Ondřej Mikoláš and Martin Bílík were responsible for image processing, while Jan Švancar and Otakar Kodytek performed software integration. "In the end, however, everyone did what was needed, including occasional sleeping," adds doc. Tomáš Haniš from the Department of Control Engineering at FEL CTU, who served as an expert advisor to the team and also provided the organisation and resources needed to implement the project. Four other students who were unable to fly to the final for financial reasons were involved in the task; Kateřina Hobzová was involved in preparing the tracks and calibrating the cameras, Ondřej Kuban worked on identifying the driving dynamics of the car, Martin Jeřábek took care of processing the car's sensor data and Jan Hadraba was involved in implementing the local planning functions.
But before the team could look forward to the finals and a trip to Toronto, they had to get through the first round of competition. It took two months of preparation, during which the students developed the system's features and implemented them in a virtual environment. For this project, the Control Engineering Department set aside a room where they could meet and work undisturbed; a printed route map, two traffic lights and a stop sign were also on hand. They also had tools from Quanser, the competition organiser, namely a digital twin of the vehicle in question. A total of 39 universities from all over the world took part in the first round of the competition, so the competition was fierce for the team from FEL CTU.
The Prague team advanced to the second round with 10 other universities, including institutions such as the University of Seoul, California Polytechnic State University and Beijing University of Technology. These teams received a physical QCar vehicle from the organizer in advance to test their algorithms before flying to Toronto. "As it happens, the team's efforts were building up as the competition approached and the team flew to Toronto," says doc. Haniš. "The real sporting enthusiasm only came during the competition days and several sleepless nights, when the whole team worked on redesigning, refining and tuning individual functions," he adds. The second round was tricky in that the vehicles were driven on a surface that the teams only became familiar with on site, which is why the competition days were really intense.
The Fast & Driverless team from FEL CTU made it to the finals as one of the three fastest teams and eventually climbed to the top position. Their concept of the control system stood out especially for its speed; the fastest vehicle in the qualification managed to cover the approximately 19 m long track in 19.8 seconds, but unfortunately the team received a two-second penalty for exceeding the track limit. However, since the time of the fastest competitors was about 25.5 seconds, the students from FEL CTU decided not to take any risks and slow down a bit in the final to the winning time of 23.472 seconds. The reliability with which the system was able to detect traffic signs was also important; the vehicle was thus able to autonomously stop before a stop sign or wait for a green light. However, the accuracy of trajectory tracking proved crucial, especially at high speeds where rivals had great difficulty.
In addition to the experience, the team took home a gold QCar trophy and a prize money, which was divided among the students. The prize also includes a QCar and a one-year subscription to QLabs Virtual QCar digital twin, which the department plans to use primarily in teaching undergraduate and graduate courses focused on control systems design.
"We entered the competition with the idea that we would try and see. Although I saw from the beginning that the students were very talented and worked on the competition with great interest, I had no comparison and I did not dare to imagine where we could place," says doc. Haniš. This was the first time the team from FEL CTU had encountered these instruments, and unlike their competitors, the students did not have much experience with them. Nevertheless, they managed to win. "That is why the first place is an absolutely great result, which the team definitely deserves," doc. Haniš praises his students.