RINA - Démonstrateur d’utilisation de l’intelligence artificielle pour la gestion opérationnelle des risques naturels gravitaires géologiques | FEREC 2020

In order to better understand the limitations and capacities of AI to predict rock falls it is important to study different sites. This variety will permit us to verify what was done on this work making sure it is applicable to different contexts. Moreover, the RN1 was a great case for a first study due to its simplicity not having the freeze thaw cycles and earthquakes, but creating models where those features are also taken into consideration could be more adapted to AI. The predictive models always aimed to predict the day a rock reached the road while another important information is the position it fell. Initial analysis of the spatial distribution was made, but no models were created incorporating the spatial aspect of the cliff. A space time model could be operational by incorporating partial closures of the road. Having sections of the road completely open helps the driver have more comfort and reduces the time of travel. 48