Virtual waterways : predictive model for fairway usage under changing conditions

This report presents a Virtual Waterways framework to simulate and predict inland fairway use under changing climate, traffic, and infrastructure conditions. It combines hydrodynamic modelling, AIS-based inverse modelling, imitation learning, and reinforcement learning to model adaptive vessel behaviour. Tested near the Hartelbrug, Botlekbrug, and Spijkenisserbrug, the framework showed that reinforcement learning can support obstacle avoidance, target reaching, and collision avoidance. Further work is needed to improve imitation learning, multi-agent behaviour, and navigation rules.