Multi-stage Simulation Framework for evaluating the impact of Autonomous Mobility on Motorways throughput and safety (VAMONA)
Founded by Croatian Science Foundation
Project ID
UIP-2025-02-1083
Project start date
15.12.2025
Project end date
14.12.2030
Abstract
The main goal of this project is to use various traffic simulation frameworks arranged in consecutive stages for systematic evaluation of Autonomous Vehicles (AVs) impact on the motorway's throughput and safety. The multi-stage evaluation process will be conducted by using different penetration rates of AVs in simulated traffic flows. Two groups of simulation frameworks will be used. The first of them is based on isolated synthetic motorway models and real-world traffic demand, while the second is based on a Digital-Twin (DT) of a real-world motorway system. Furthermore, a secondary goal is to use the proposed stages of simulation frameworks to train one driving aspect of AVs with a strategic goal of improving overall motorway throughput. It will serve as a use case for proposed multi-stage evaluation ecosystems based on traffic simulations. Thus, additional traffic scenarios will be created by generative AI models for synthetic motorway models to boost the variety of simulated traffic scenarios. This approach will potentially improve the training convergence and simultaneously introduce a more comprehensive evaluation of AVs at the early stages of the proposed multi-simulation evaluation ecosystem. Finally, the DT will be used for long-term simulation evaluation of trained AVs using real-time traffic data on real-world motorway systems. The DT can be used for retraining the AVs if the first evaluation provides poor results regarding the overall throughput and safety.
Keywords
Autonomous vehicles, generative artificial intelligence, microscopic traffic simulators, digital twin, motorway traffic control
Project members
Project leader: Asst. prof. Martin Gregurić, PhD.,
Lucija Bukvić, PhD.,
Krešimir Kušić, PhD.,
Filip Vrbanić, PhD.,