The issue of how to tackle traffic congestion is one that will be forefront in the thoughts of any transport planner in major cities around the world. As the number of vehicles on the roads grows, new solutions and technologies will be needed to ensure jams do not bring cities to a halt.
One solution could be artificial intelligence. Researchers at Nanyang Technological University in Singapore have developed a new intelligent routing system that aims to minimise the number of spontaneous traffic jams – those that are not caused by accidents or roadworks, but those that are simply a result of too many cars trying to share the same space – by rerouting some traffic.
Vice reports the algorithm works by assessing current traffic loads in real time and identifying where problems are most likely to occur. It can then give drivers recommendations on the best paths to take in order to minimise the risk of traffic flow breaking down into congestion.
While route-finding is a commonly-used branch of computer science, one of the key challenges involved in creating a traffic algorithm is that even small-scale deployments with just a few possible paths can quickly become hugely complex.
To analyse traffic across an entire city therefore involves a massive number of potential routes and factors – both predictable ones such as traffic light patterns and unpredictable ones such as individual driving styles.
Presenting the research in the April edition of the IEEE Transactions on Emerging Topics in Computational Intelligence journal, the team explain that to handle this, the algorithm recognises that eliminating traffic jams altogether is not possible. Once the system has an inbuilt assumption that breakdowns in flow are going to occur, it becomes a case of how these can be minimised to have the least impact.
“Our goal is to direct the traffic flow so that the overall traffic breakdown probability is minimised,” the researchers wrote.
One of the key findings to come out of the research is that the technology does not not need to optimise the route-finding for every vehicle in order to be effective. In fact, it claims that just ten per cent of vehicles in a network need to be connected to and following the routing optimisations to see a positive impact across the entire network.