guest ::
TY - THES AU - Woopen, Timo TI - Kollektives Verkehrsgedächtnis zur Unterstützung vernetzter automatisierter Verkehrsteilnehmer PB - Rheinisch-Westfälische Technische Hochschule Aachen VL - Dissertation CY - Aachen M1 - RWTH-2025-06480 SP - 1 Online-Ressource : Illustrationen PY - 2025 N1 - Veröffentlicht auf dem Publikationsserver der RWTH Aachen University N1 - Dissertation, Rheinisch-Westfälische Technische Hochschule Aachen, 2025 AB - „Learning is experience. Everything else is just information.“ This quote by Albert Einstein also applies to mobility, which is undergoing one of the greatest revolutions since the invention of the internal combustion engine. Connected and automated driving promises to make future mobility significantly safer and more comfortable. Megatrends such as artificial intelligence and connectivity enable centralized data collection and subsequent learning to continuously improve implemented functions. Humans, on their part, learn safe and comfortable driving through their own experiences and by implicitly evaluating their own and others’ maneuvers in their environment, thus building location- and time-coded (spatio-temporal) commuters’ knowledge. However, automated driving functions are usually developed and optimized for the entire application. The collection and generation of location- and time-specific experiences hold the potential to especially support connected and automated vehicles through their own or others’ experiences, similar to the human learning process. Therefore, this work conducts a research investigation and prototypical development of a collective traffic memory system to support connected automated vehicles by means of the aggregation of spatio-temporal fleet knowledge.To this end, it is investigated how fleet knowledge can be built and the experience contained therein may be described. For this purpose, evaluation aspects and metrics are introduced and combined to a posteriori derive experience from trajectories encountered situations. This approach considers that experience for automated vehicles can be aggregated not only from their own experiences but also from those of a fleet or from additional infrastructure data. A cloud architecture is developed to collect events and then derive relevant experience. It can process data from different vehicles and other sources in real time, store it, and make it available for further use by different applications based on analysis rules. This architecture serves as the foundation for the exemplary implementation of an application to generate trajectory suggestions at urban intersections using the collected spatio-temporal experience.The results show that trajectories can be evaluated a posteriori by combining appropriate metrics at different evaluation aspects to derive experience. This experience can be expressed using comparable parameters and can be utilized in various applications. Compared to a conventional trajectory planner, the usage of experience-based trajectory suggestion shows visible positive effects on the comfort and safety of the planned and executed trajectory during normal drives. Furthermore, the use of spatio-temporal experience shows great potential in handling situations arising from temporal or spatial changes in the traffic environment, such as temporary construction sites. LB - PUB:(DE-HGF)11 DO - DOI:10.18154/RWTH-2025-06480 UR - https://publications.rwth-aachen.de/record/1015619 ER -