The SWISSTRAFFIC Group conducted an in-depth analysis of an existing complex junction for the Vosges Departmental Council, which had been causing safety problems when entering and exiting the junction. The objective was to identify potential safety concerns and propose effective redesign measures. Additionally, an accurate count, classification, and origin/destination analysis of mobility were crucial to better understand the traffic dynamics within the junction.

To meet the requirements, the SWISSTRAFFIC Group introduced its innovative solution swissTRAFFIC AI – an autonomous pole-mounted camera with in-built AI processing. The swissSCOUT AI is able to autonomously process a whole week of real-time video, monitor the system remotely, and automatically retrieve the analysed data. The swissSCOUT AI’s artificial intelligence enables counting and classification into 14 categories, with accurate lane tracking at the level of a large junction and warning of near-miss risk areas. The assessment of the risk of near misses is based on automated AI recognition of potential conflicts between directional movements on the road.

The study transferred the entire Origin/Destination (O/D) matrix, which was classified in 15-minute intervals, as well as the matrix of near misses, which reported several risk situations throughout the week of analysis. Artificial intelligence also made it possible to visualise the trajectories at the intersection in the form of a heat map.

Compared to studies conducted with investigators or other technologies, the Artificial Intelligence used by SWISSTRAFFIC Group enables more detailed surveys of counts, instantaneous speeds, and source/destination over long and more representative periods of time, day and night, without the need for external lighting. The data collected allows modelling software to be fed with accurate data. Artificial intelligence is a powerful tool in recognising near misses to take concrete and reasoned action.