From Cameras to Insights: VisionAI Transforms Motorway Operations

Vehicle counting, speed of traffic, congestion, illegal turns, safety… vision AI has become one of the key technologies in mobility due to its ability to detect virtually any event happening on roads, which is transforming how cities and operators understand movement. This didn’t go unnoticed to Attikes Diadromes, the Greek company that operates, maintains, and manages Attiki Odos, the main toll highway system in the greater Athens area, a motorway with a length of 70km and an average 210,000 daily entries. As part of their concession contract, Attikes Diadromes has had to perform major maintenance works, something tricky considering any works could disrupt traffic. To avoid this issue, the company needed to know exactly when and where closures could happen for minimal impact. That traffic (data) is very interesting for us in two ways. First of all, to know what is the traffic and what is the speed in different areas along the 70 kilometer motorway that we have. And secondly, if you know the traffic and you know the pattern of the traffic, you know where to interrupt this traffic or if it can be interrupted to do maintenance work. To understand mobility patterns, the company turned to AI software company Goodvision in 2023. The startup’s agnostic AI platform analyzes traffic footage 24/7 from any camera. Attikes Diadromes already had a CCTV network, so they connected those cameras to the platform, which can run on servers or on the edge. Once deployed, the platform starts analyzing footage and sending alerts into traffic management systems. The motorway operator started by testing the software on four cameras and quickly expanded to twelve across the 70-kilometer stretch once they saw the value, a number that can continue to grow as needed. They’re using the platform to recognize traffic patterns in real time and plan maintenance precisely at the least disruptive moments. Before, what we had, what we used in the market are like heatmaps, which means that you had to develop from all your loops a map for October-November, that was showing, if you wanted to do work in October, the previous year, what are the hot points within. Here we had instead, a real time knowledge through the AI cameras. So do you want to look backwards two years or three years or a set of years to understand if I don't interrupt? Although Attikes Diadromes is currently using vision AI primarily to optimize maintenance operations, the technology opens the door to far more: classifying vehicle types, identifying near-miss events, spotting violations, optimizing traffic signals, analyzing movement patterns to improve overall road safety... vision AI has virtually endless use cases, limited only by how the system is trained. You know, very often it happens that our customers are actually defining the use cases and applications and we are sometimes very impressed, how many take tens and hundreds of applications that the system can have. So in this case, there was a combination of basic traffic flow, performance parameters monitoring, combined with an incident detection or event detection on the roads that together resulted in a very powerful information system. vision AI is becoming one of road operators’ greatest allies. Turning camera feeds into insights allows them to focus their resources exactly where needed, when needed. As the tech adapts and continues to expand its reach, we’re bound to see more operators and cities bet on vision AI. Ioannis, who is currently working on other road projects, even mentioned he wants to use AI again, which is only a good sign for the industry.