How can AI be used in the smart city of the future to significantly improve the performance and reliability of future fiber optic networks at peak times, as well as their cost and energy efficiency? This question was answered over a period of three years by the KIGLIS project (Artificial Intelligence for the Optimization of Fiber Optic Networks in a Smart City), which was mainly funded by the Federal Ministry of Education and Research with a total of four million euros. In mid-October 2023, the Karlsruhe Research Center for Information Technology FZI presented the support of autonomous driving via remote assistance in the Baden-Württemberg autonomous driving test field. The presentation was part of the final event of the KIGLIS research project as part of the KIT Science Week.

Unterstützung des autonomen Fahrens in komplexen Verkehrssituationen

Autonome Fahrzeuge werden auch in naher Zukunft immer wieder menschliche Unterstützung benötigen, wenn komplexe Situationen auftreten. Diese Unterstützung muss aber nicht zwingend vor Ort präsent sein: Mittels Remote Assistance könnten auch Menschen in Leitstellen aus der Ferne dem autonomen Fahrzeug aus diesen komplexen Situationen heraushelfen. Im Rahmen der oben genannten Abschlussveranstaltung des Forschungsprojektes KIGLIS wurde die echtzeitfähige Anbindung von autonomen Fahrzeugen zur entfernten Unterstützung erfolgreich demonstriert.

New methods for reliable data transmission

For the support of autonomous vehicles by a control center, for example to ensure the constant availability of a vehicle fleet, the reliable transmission of a large amount of sensor data from the vehicle and, if necessary, from the infrastructure sensors is indispensable. In order to guarantee this in a future smart city with many networked units and thus a high volume of communication, the FZI has researched and developed various methods specifically for autonomous driving in the KIGLIS project. These include, for example, AI-based compression methods for LiDAR and camera data. These new methods, together with the partners' network infrastructure, have now successfully demonstrated their performance on the autonomous driving test field. In addition, the FZI researched methods for detecting so-called corner cases, i.e. very rare traffic situations that also require remote assistance.

Strong partners

In addition to the FZI Research Center for Information Technology, the joint partners also included the Karlsruhe Institute of Technology (KIT), Nokia Solutions and Networks GmbH & Co. KG, VPIphotonics GmbH, Atesio GmbH and TelemaxX Telekommunikation GmbH.

Video