Artificial intelligence for track limits tested in Vallelunga

The world of motorsport deals almost every week with the track limit problem and the now usual accompanying controversies, also given the complexity of monitoring everything the drivers do on the track.

Just think, for example, of what happened in the Austrian Formula 1 Grand Prix in 2023, when the Federation was forced to verify over 1200 cases during the race, with a notable delay in assigning penalties. This clearly caused a lot of discontent at the time, also because some drivers had not received any warning due to this delay, as usually happens, directly receiving a penalty without being able to react.

To be able to monitor possible infringements with greater accuracy and speed, help is sought from the world of technology, in particular from artificial intelligence linked to Computer Vision. The idea is that this system is used to examine each pixel in a certain area within the video, so that we can judge whether cars have exceeded the limits of the track. Above all, this new approach should allow it to be much more effective in reducing the number of possible infringements that require a final check by a human, thus speeding up the process.

The first Formula 1 tests were carried out at the end of last year in Abu Dhabi on an experimental basis, but different solutions and systems are being tested around the world. An example is the test carried out at the Piero Taruffi Autodromo in Vallelunga in recent days, in which a track limit management system developed by Mermec Engineering (a company part of Angel Holding, led by the President Cav. Vito Pertosa, which develops solutions for the railway, digital mechatronics and aerospace sectors).

Explaining more about how the system works was Niccolò Chierroni, CEO of Mermec Engineering, who illustrated how the experimentation started last year in the world of two wheels, and then expanded to four wheels through ACI. “The idea was born from the interaction with the Race Direction, which then finds itself during the races having to manage many track limit reports, which are often not clearly objectified. This therefore gives rise to disputes and management problems.”

“We therefore thought that our automatic video analysis systems, therefore artificial intelligence applied to computer vision, could be of support. The system is therefore able to automatically recognize the cars on the track, monitoring when they cross a limit line in the curves pay attention. Following the identification of the infringement, the Race Direction receives an extract of the video in real time, combined with a history of the information, thus being able to proceed with the possible application of sanctions”.

The most interesting aspect mentioned by Chierroni is that this system could already rely on the existing camera circuit, although clearly more recent and high-resolution instruments could provide even more effective detections: “A simple and intuitive system which, which can symbiotically connect to the cameras already present in the circuit and provides precision in the order of centimetres, mainly dependent on the resolution of the cameras installed. Versatility is also among the strong points, with the “limit lines” easily adaptable to any type of curve”.

The response of the President of the ACI was positive, immediately after taking part in the presentation: “The impression is excellent: an innovative and reliable technology. I believe that exploiting certain opportunities offered by artificial intelligence is fundamental, also to best respond to the challenges of the future.”

This is undoubtedly a solution that can represent the future of motorsport in the management of track limits, especially considering that, for example, in Formula 1 the FIA ​​was forced to remove the sensors because they proved to be too careless. In fact, the loops also reported erroneous reports, which were later verified by the Federation inspectors, which is why these sensors were then deactivated in view of the transition to artificial intelligence.

“Essentially, what we concluded after Austria was that all three data sources were sending potential reports and that if you excluded all those incorrectly reported by the loops and analyzed whether all those correctly reported by the loops were also captured from the man, the answer was yes [vi era un track limit]”, Tim Malyon, deputy race director and head of the FIA ​​Remote Operations Center in Geneva, explained at the end of 2023.

“So we concluded that the loops were not accurate enough. And that by far the most accurate solution was to have a data analyst look at the video itself.”