The Audi Autonomous Driving Cup

The Audi Autonomous Driving Cup is an interdisciplinary competition focusing on students of Computer Science, Electrical Engineering and related disciplines. The students develop automated driving functionalities on miniature vehicles facing similar challenges as on full-size cars. The miniature vehicles as well as the basic driving functionalities are delivered by Audi, providing a perfect basis for developing higher-level perception and planning algorithms.

The interdisciplinary character as well as the development of industry-related topics are the perfect reason for the FZI to support this competition. The idea of developing real­-world applications on small­-scale vehicles was highly integrated into several projects. Over the last years, a supporting infrastructure has been built up including several small­-scale environment objects of different qualities and quantities as well as a full­-featured intuitive simulation for the efficient prototype testing of the algorithms.


Team AlpaKa 2018

Team AlpaKa, supported by the FZI, was one of 16 teams participating in the Audi Autonomous Driving Cup 2018. The students Simon Roesler, Maximilian Zipfl, Shuxiao Ding, Mark Timon Hüneberg and Yimeng Zhu celebrated the first victory of the FZI at the AADC.


Team AlpaKa built on the experience and software of Team KADABRA and extended it, in order to meet the extended requirements of AADC 2018. The newly introduced challenges, such as driving over bridges and giving priority to rescue vehicles, were mastered by Team AlpaKa, just as the "Free Challenge" mode, in which the team presented a distributed overall system for tomorrow's local traffic. In addition to autonomous shuttle operation for extended mobility with dynamic and adaptive route planning based on the current congestion situation, the team also developed a teleoperation of the vehicles used for possible malfunctions, in which a human can intervene as the last instance. Team AlpaKa convinced the jury with its detailed and conclusive overall concept and its implementation.

To the press release


 

Team KADABRA 2017

Team KADABRA, supported by the FZI, was one of the ten teams that participated in the third Audi Autonomous Driving Cup 2017. The students Robin Andlauer, Kolja Esders, Nico Kuhn, Fabian Dürr and Kai Braun  could unfortunately not defend the second place of the years before. Team KADABRA built up on the experiences of the teams KAtana and KACADU. The software architecture was altered from a monolith towards a modularised micro-architecture. An enhanced robust road perception was achieved by means of Fully Convolutional Neural Networks. The necessary training data was generated in simulations, which allows for automated labelling. Furthermore, a new method to detect and track cars and pedestrians was implemented by incorporating the YOLO real-time object detection system.


 

Team KACADU 2016

Team KACADU, supported by the FZI, was one of the ten teams that participated in the second Audi Autonomous Driving Cup 2016. The students Vitali Kaiser, Jan­-Markus Gomer, Micha Pfeiffer, Peter Zimmer und David Zimmerer were able to defend the second place from 2015.

Team KACADU built up on the experiences of team KAtana. The modular road environment was enlarged by a few high-quality modules, for realistic competitive conditions, and large, quickly reconfigurable carpets, which enable an extensive testing with various vehicles. The idea of the robust hierarchical road layout perception was kept up and completely re-implemented in a more performant grid­-based solution. In a parallel lab course, students from KIT laid the basis for the miniature vehicle simulation, which will be expanded and extended for future competitions

Further information (in German)

 

Publication

Details about the robust environment perception were presented in the framework of a paper on Intelligent Transportation Systems from the 2016 IEEE International Conference by the research team on 1­4 November 2016.

The publication can be found at: https://www.researchgate.net/publication/309675872_Robust_Environment_Perception_for_the_Audi_Autonomous_Driving_Cup

Source Code

The source code is available for download at GitHub: https://github.com/fzi-forschungszentrum-informatik/aadc2016

Team KAtana 2015

Team KAtana, supported by the FZI, was one of the ten teams that participated in the first Audi Autonomous Driving Cup 2015. The students Raphael Frisch, Christoph Rist, Nilan Marktanner, Philipp Hertweck, Sascha Ibrahimpasic made it into the finals and were ranked second.

Team KAtana was the FZI's first team participating in the cup. They faced several challenges and laid the basis for future teams. The first physical road modules were created and a first implementation of the robust hierarchical road layout perception was used in the cup. Furthermore, the necessity of a supporting simulation was identified.

Further information (in German)

 

The source code can be downloaded at GitHub: https://github.com/KAtana-Karlsruhe/AADC_2015_KAtana

Simulation environment for highly-automated miniature vehicles

For the Audi Autonomous Driving Cup, the FZI has developed a simulation environment for highly-automated driving functionalities of miniature vehicles, based on the open source environment Gazebo. It consists of

  • an intuitive scenario editor, including models for road elements and traffic signs,
  • a physical and visual vehicle model of the miniature vehicles, including all actuators and sensors
  • behaviour models for elementary simulations of other traffic participants

Publication

Details about the underlying concept of the simulation are presented in the framework of the IEEE International Conference on Simulation, Modelling, and Programming for Autonomous Robots (SIMPAR) on December 16, 2016.


Contact

Dr-Ing. Florian Kuhnt

Senior Expert

more

Career

Dr.-Ing. Florian Kuhnt ist Senior Expert für Automatisiertes Fahren und Situationsverstehen. 2020 promovierte er am Karlsruher Institut für Technologie (KIT) zum Thema „Holistic Temporal Situation Interpretation for Traffic Participant Prediction“. Er arbeitet als Projektleiter und wissenschaftlicher Koordinator in zahlreichen deutschen und internationalen Forschungsprojekten sowie Industriekooperationen, unter anderem:

 

Publications

zu den Publikationen

Contact

Phone: +49 721 9654-364
E-Mail: kuhnt@dont-want-spam.fzi.de

hide details

Tobias Fleck (M.Sc.)

Wissenschaftlicher Mitarbeiter

more

Career

Tobias Fleck studierte bis 2016 Informatik am Karlsruher Institut für Technologie (KIT). Während seines Masterstudiums beschäftigte er sich mit Kognitiven Systemen, insbesondere probabilistische Schätz- und Modellierungsmethoden bildeten seine Schwerpunkte.

Seine Masterarbeit "Trajectory Estimation and Prediction in the Context of Autonomous Driving" führte er in der Abteilung Technisch Kognitive Systeme (TKS) durch.

Seit Februar 2017 ist Tobias Fleck als wissenschaftlicher Mitarbeiter am FZI Forschungszentrum Informatik in der Abteilung TKS tätig.

Publications

zu den Publikationen

Contact

Phone: +49 721 9654-216
E-Mail: Tobias.Fleck@dont-want-spam.fzi.de

hide details