KI Wissen

KI Wissen supports the ROAD-R 2023 Challenge

Road Event Detection with Requirements – Call for Participation

The KI Wissen project makes the ROAD-R competition possible by supporting it with knowledge, data and sponsorship. Researchers, students and practioners with the focus on machine learning are invited to participate in the challenge. 

After successful competitions in the past years, ROAD-R 2023 started in June, supported among others by the KI Wissen project partner Fraunhofer FOKUS.

The ROAD-R challenge has been officially accepted at NeurIPS 2023 Conference and will be part of the NeurIPS2023 Program in December in New Orleans, Louisiana, USA. The NeurIPS (Neural Information Processing Systems Foundation) is a non-profit corporation whose purpose is to foster the exchange of research advances in Artificial Intelligence and Machine Learning, by hosting an annual interdisciplinary academic conference. (

The current challenge is built around ROAD-R, an extension of the ROad event Awareness in Autonomous Driving (ROAD) Dataset which is based on (a fraction of) the Oxford RobotCar Dataset. ROAD-R is the first real-world dataset equipped with requirements written in propositional logic that capture background knowledge on autonomous driving. (

Call for Participation

Short description of the challenge
The ROAD-R Challenge provides participants with a unique dataset consisting of 22 long videos annotated with road events. Further, a set of 243 requirements expressing hard facts about the world is provided (e.g., "a traffic light cannot be red and green simultaneously”). The challenge consists of two tasks, and participants can take on either of them, or both!

  • In the 1st task, participants have to develop the best-performing model using only a subset of the annotated data. This encourages leveraging the background knowledge expressed by the requirements to facilitate training on the unlabelled portion of the dataset.
  • In the 2nd task, participants have to create systems whose predictions adhere to the provided requirements. This presents a fascinating challenge where AI systems must not only detect road events accurately but also ensure compliance with human-level requirements.

Important Dates
Challenges open for registration: June 13, 2023
Training and validation data release: June 13, 2023
Test data release: August 15, 2023
Deadline for results submission: September 30, 2023
Announcement of results: October 7, 2023