DC12. Harmonisation and hybrid application of AI datasets for road safety

Hosts: FPZ & iRAP

Objectives:

• To create a taxonomy for classifying datasets generated by AI algorithms applied to data from various sources (e.g., Lidar, 360 imagery, geo-spatial, telematics, satellite imagery and traffic management) and assess their transferability
• To create a new methodology for harmonising datasets generated by both AI algorithms and manual approaches
• To develop new hybrid road safety assessment models that effectively combine the advantages of various data sources, for determining the road crash occurrence risks and the risks of serious injuries to motorised road users and VRUs

Expected results:

• Knowledge and methodologies for the alignment and harmonisation of datasets generated by AI algorithms from various sources
• New AI models for extracting road and traffic information from various geo-referenced sources and their correlation with risk rates
• Application of the methodology in an appropriate case study or application compatible with the mission and activities of iRAP

Planned secondment(s):

EIRA-SI, Purpose: to test the developed methods on additional datasets