Going with the flow

The fight against congestion is getting some new tools: mobile phones and complex algorithms.

BY Edwin Cartlidge
4 April 2016
FEATURESMobilityUrban mobility

The take-away

  • Thanks to smartphones, planners are collecting data that will help improve traffic flows.
  • Equally important is providing drivers with information that allows them to make better decisions.

Gridlocked streets, shortage of parking spaces, pollution, road rage and, unfortunately, accidents. These are familiar problems to anyone living in cities, and they are set to get worse as more and more people flock to urban areas. The response of cities such as London and Stockholm has been to charge a fee to enter the central area. Others, including Copenhagen, are trying a different solution: using modern computing and communication technology to improve the flow of existing traffic.

The key tool for Copenhagen is the smartphone. In 2013, the city council commissioned a group of experts from the Technical University of Denmark (DTU) and various companies to look at the feasibility of monitoring traffic flow via a network of Wi-Fi “access points” across the city. The idea was to exploit the fact that most people on the move – be they motorists, cyclists, public-transport passengers or pedestrians – now carry a smartphone. Using signals that the phones continually emit, the system would calculate the person’s position and speed.

To test the idea, the researchers set up six access points on lamp posts along a busy stretch of H.C. Andersens Boulevard. By comparing data from the access points with signals from GPS units given to a number of cyclists and pedestrians, the experts were able to track individuals to within a metre of their true position. They were also able to convince lawyers at the town hall that the procedure wouldn’t violate privacy legislation, by stripping data of the numbers that identify individual phones.

According to DTU’s Per Høeg, these data could have many uses. One would be monitoring the density of cyclists at key junctions in order to advise individuals to change route or slow down when necessary, particularly at rush hour. Cyclists, he says, can be “very pushy”, causing some to fall close to passing cars, and they often ignore red lights. “Every year someone is killed,” he says.

In addition to monitoring parking spaces and turning street lamps on and off for pedestrians at night, perhaps the main benefit of the new data will be in managing traffic lights. Currently “green time” – the duration of a green traffic light – is controlled by a computer program that processes images from roadside cameras. But Høeg says the program can’t discern individual vehicles when the traffic gets too heavy. “At that point you need to set the green time manually.”

The need for green

Finding ways to make best use of green time is a key aim of many researchers in Europe and beyond. In Germany, a collaboration between universities, research institutes and industrial companies known as UR:BAN has developed technology to minimize the number of trucks that stop at red lights. Since trucks take disproportionately longer to start up again than the equivalent length of cars, platoons” of trucks travelling not far behind one another are allowed, if possible, to travel through each junction as a single unit – thereby improving traffic flow while reducing pollution and noise.

The system uses roadside detectors to pick out trucks from among the vehicles approaching a set of traffic lights, and a computer program to work out whether to keep the lights green for an extra few seconds. The program models traffic building up behind the red lights to establish whether the extra green time will create a net benefit.

Fritz Busch of the Technical University of Munich (TUM), one Simulating cities of the project partners, says that a trial involving a single test truck in Düsseldorf shows that the scheme is “technically feasible”, but he admits that it remains unproven in a “statistically significant way”. He adds that in future drivers should also receive advice about the speed needed to keep platoons as compact as possible. He points out, however, that the system will need to be standardised across cities and countries if truck manufacturers are to incorporate it into their vehicles.

Making the best use of green time is also the aim of researchers at Newcastle University, who are taking part in a European Union project known as Compass4D. The group has developed a device that tells drivers how fast they should travel to hit the next traffic lights on green and that allows them to change lights to green if necessary. The device is being trialled in a number of non-emergency ambulances in Newcastle and, if it improves traffic flow, could be fitted to trucks, buses and taxis as well.

Opening up the open road

A quite different way of optimising green time is being taken by Nikolas Geroliminis and colleagues at the École Polytechnique Fédérale in Lausanne (EPFL). The Swiss group is analysing entire road networks rather than specific intersections. Instead of modelling the position and speed of individual vehicles, which requires huge amounts of data, they monitor the number and average speed of vehicles across the network. “We try to operate traffic in cities in a holistic way,” he says.

The researchers have developed an algorithm that coordinates a network’s traffic lights so as to maximise the total distance travelled by vehicles in the network every minute. Geroliminis explains that this “flow” increases as a city fills up, but that at a certain point average speed slumps to such an extent that the flow starts to drop. Eventually, the network becomes gridlocked and flow ceases. We want traffic levels to be neither too low nor too high,” he says.

The group has already simulated traffic flow in San Francisco and Barcelona, finding that its algorithm should reduce congestion by more than 20%. Now it is moving into the real world, developing “smart traffic lights” that will be controlled by its algorithm during tests planned for Geneva in 2017. While computer simulations are a powerful tool, says Geroliminis, “you can’t always simulate individual people and traffic accurately”.

It is not just in city centres that modern technology is being used to improve traffic flow. Currently a long stretch of motorway running from Rotterdam, via Frankfurt, to Vienna is being fitted with Wi-Fi access points to enable drivers and traffic control centres to communicate with one another. Drivers will be warned about upcoming dangers such as slow-moving road works, while controllers will be automatically informed of vehicles’ positions and speed. The project got underway this year, and, according to Busch, could be expanded to include all German motorways by the end of 2018.

Improving the information that drivers receive is also one of the aims of the UR:BAN initiative. Researchers are developing a windscreen-integrated display that allows drivers to keep their eyes on the road, as well as “active gas pedals” that become less responsive when drivers need to be alerted. They are also simulating drivers’ behaviour so that cars can provide more meaningful warnings, and are optimising emergency braking and other automatic interventions to strike the right balance between driver autonomy and safety.

As well as improving safety, these technologies should contribute to the broader aim of improving traffic management. But, as experience in Copenhagen shows, trying to get traffic flowing more smoothly is easier in theory than in practice. Problems with power supply cut the number of access points that could be used for the smartphone trials from a planned 20 to six, and those that remained happened to be on lamp posts next to trees whose leaves reflected wireless signals differently when wet and dry. A shorter than expected sampling distance – 200 m instead of 1 km – also made it difficult to track fast cars. “We had lots of little nitty-gritty problems that we had to overcome,” says Høeg.

Such teething problems aside, he believes that modern technology is essential in providing relatively cheap and efficient ways of making cities safer and more environmentally friendly. Busch also thinks that in the long run automated traffic is inevitable, particularly if city planners want to avoid building ever more roads. “Smart systems are definitely the future of traffic,” he says.

Simulating cities


How do people travel and where do they go? Models that simulate individual travel behaviour can help answer these questions to help design better cities. These so-called activity based models predict which activities are conducted when, where and with what mode of transport. Diverse parameters that influence an individual’s travel decisions are taken into account, such as number of children or choice of work and workplace. This high level of precision allows for forecasts of traffic’s impact on air pollution.

One such model, Albatross, was developed by Harry Timmermans at the Eindhoven University of Technology under commission from the Dutch Ministry of Transport. According to Mobility and Traffic Professor Timmermans, this research is crucial, especially in the Netherlands, where “transportation and environmental planning are prominent due to a lack of urban space”.