Tunnels are amongst the most demanding environments on Europe’s road network. Confined spaces, limited visibility and the difficulty of intervening quickly in an emergency all raise the stakes when something goes wrong. Across the MATIS programme, 3 partners have each tackled this challenge and show how MATIS partners are approaching the same underlying question — how to make tunnels safer — from genuinely complementary angles.
Cerchiarello: equipping the tunnel itself
On the A4 Torino–Milano motorway, SATAP focused on the Cerchiarello tunnel, an 800-metre twin-bore structure sitting at a complex interchange node. The tunnel was equipped with a comprehensive package of ITS devices: 18 AID (Automatic Incident Detection) cameras, 7 cross/arrow signs, 3 variable message signs, 4 entrance traffic lights, 127 lighting devices (15 of which are dedicated blue safety-distance lighting), and 5 SOS stations complete with LED signalling and radio retransmission cabling.
The aim is straightforward: reduce the number of accidents and cut the time it takes to manage an emergency once one occurs. The project also included configuration and upgrade work at the control centre itself, so that the new equipment integrates properly with existing operations.
traffix.ai: teaching the cameras to think
Cameras alone are only as good as the software interpreting what they see. That was the starting point for SINA’s work with CDT on the A12 Sestri Levante–Livorno and A15 La Spezia motorways in Tuscany. The issue wasn’t a lack of detection technology: it was the sheer number of false alarms it generated. A change in brightness, or something as trivial as a sheet of paper blown onto the sidewalk, could be misread as a pedestrian, triggering an alert that operators then had to investigate.
The solution, deployed across roughly 26 tunnels, is the traffix.ai software: a system combining 3D object tracking with a Deep Learning module to classify vehicles and detect abnormal traffic situations in real time. The results speak for themselves: in the May–September 2024 evaluation period, the average false alarm rate fell to just 3%, down from around 90% before the software was introduced. That is a dramatic reduction in the noise operators must filter through, freeing them to respond faster to events that are genuinely happening.
AUCAT: a full-scale overhaul on the C-32
On the C-32 motorway near Barcelona, AUCAT faced a different starting point altogether: not one tunnel, but sixteen, spread across a 47 km coastal corridor carrying more than 26,000 vehicles a day. Rather than a single fix, the project set out to modernise the whole safety chain at once, tying every upgrade together through the HORUS tunnel management platform.
Optical sensors on roads and ramps detect oversized trucks before they reach a tunnel too low for them, triggering automatic alerts on nearby signs to guide them off safely. Over 100 new LED warning signs, built to withstand the weather, have replaced faded older ones, displaying clear day-and-night messages on hazards and speed limits. Slow mobile links have been swapped for dedicated fibre connections, giving signs and equipment an instant, reliable line to the HORUS control room. Inside the tunnels themselves, cameras now monitor escape routes, lane lights guide or block traffic at entrances, and backup power keeps everything running through outages, all coordinated by smart controllers reporting into the same platform.
The results reported so far are substantial. Detection times have improved by around 12 seconds, keeping more vehicles from entering tunnels where an incident is already unfolding. Queues caused by incidents have been cut by roughly 50%, freeing up more than 10,000 vehicle-hours a year and accident risk inside the tunnels is expected to fall by around 1.5 crashes annually.