An Israeli firm will in 2022 deploy its AI-based software for preventing traffic jams before they form at a junction in Luxembourg, the location for which is yet to be announced.
ITC.city, which was among the laureates of Luxembourg’s twelfth Fit 4 Start acceleration programme, is currently in talks with local authorities to demonstrate its software by integrating it with existing intersection cameras and traffic lights.
“Traffic is a problem that is not going to disappear,” Dvir Kenig, founder and CTO told Silicon Luxembourg. “The traffic jam is one of the biggest problems and the range of solutions is very limited. Once we have technology that is proven to work, we will succeed in reducing traffic jams by 30%.”
Predictive management
Congestion is a major contributing factor of greenhouse gas emissions. Even if countries succeed in electrifying their pool of vehicles, there is still the issue of time lost in traffic. According to data published by mobility geolocalisation firm TomTom, Luxembourg City was the 106th most-congested city in the world in 2020, with motorists losing up to 105 hours in rush-hour traffic per year. Yet the situation in the grand duchy’s capital is less painful than in Tel Aviv, the city where Dvir and his team first began, which TomTom ranked as the 25th most-congested city in 2020.
For ITC.city’s team, which includes Dvir’s machine learning professor, the AI entrepreneur Professor Amir Geva, this local problem gave an impetus to develop a predictive management solution for traffic lights at junctions. “Predictive management says that even if now in the south there is less traffic, I will still give more time to managing traffic in the north because 10 minutes from now, there will be a lot of cars that will come there,” the CTO explains.
400 parameters per second
Currently, traffic lights operate on a relatively simple computer system fed by traffic data from sensors in a given location. But, the entrepreneur says that what is lacking is a mathematical model that can give a behavioural overview over a longer period of time for all of the different junctions and then manage lights dynamically to reduce jams. ITC.city says that its solution considers up to 400 parameters per second, generating a rich and GDPR-compliant data set on a range of road users that would be impossible for the human mind to process.
“If you give this to machines that can process these kinds of things, it will surpass the human ability to manage traffic. And this is the data-driven approach,” says Dvir.
In Tel Aviv, ITC.city has tested its solution on one of the city’s business junctions, Namir and Einstein, where it manages over 170,000 vehicles daily. In the first half of 2022, the firm will expand the solution to five more junctions.
A Luxembourg office
But the market in the world’s unicorn capital is limited compared to Europe. Once the pandemic air travel situation stabilises, Dvir plans to open an office in Luxembourg where he will deploy the tech firm’s software at a single junction, as part of a European pilot. He hopes that this proof of concept will provide a launchpad to expand into the largest European markets of France and Germany.
Another attraction for Luxembourg, beside its tech-friendly eco-system, is MeluXina, the 10-petaflop super computer inaugurated in 2021, which Dvir says could enhance the potential of the AI technology behind the solution to the point it would be like going from driving a Mercedes to a Ferrari.
With the HPC, “we can do wonderful projects. If this were possible, it would allow us to do something new, work that has never been done before. No-one in the world has used an HPC to manage traffic,” says Dvir.
