The infrared sensors that could make cycling safer

In the UK’s major cities, more people than ever before are hopping on their bikes and cycling to work. The number is up 144% over the past decade in central London, and more than 80% in Brighton, Bristol and Manchester, according to the Office for National Statistics.

With more cyclists on the road, it’s important to figure out how to give them space to be safe and get around efficiently, without cutting into pedestrian walkways or forcing them to do battle with cars.

Transport charity Sustrans is hoping to provide a solution, and it begins with finding out how many cyclists are actually on the road. The charity will then use the data collected to improve cycle lanes and other road infrastructure.

“We wanted something that could tell us how many people are using the space now, so we can adapt the streets to make them better for walking and cycling,” explained Dr Cecilia Oram, project officer at Sustrans.

New York’s Times Square faced a similar challenge. “There were quite a lot of incidences of people walking in the carriageway, and therefore there was quite a high rate of accidents,” she explained. “They explored how they could model the junctions to make it easier for pedestrians to get across the road and make more space in the carriageway for them.”

Counting cars and bikes

Simple, right? Not so. While you can easily count cars with a smart strip in the road or via a camera, cyclists and pedestrians are harder to identify. You could use a standard camera and develop an algorithm to distinguish pedestrian from cyclist and understand their movement, but this raises privacy and dataprotection issues because individuals can be identified. Indeed, Oram’s initial solution was to use CCTV, but privacy concerns put paid to that idea.

Sustrans’ new plans make use of a Dutch- and Danish-developed infrared sensor system that can count people without raising data-protection fears. “It’s anonymous data; you can’t recognise people from the thermal image,” she said.

In addition, it works regardless of the conditions, noted Rikke Gade, an assistant professor at Aalborg University in Denmark, who is developing the system. “It doesn’t rely on light at all, so it works both day and night,” she said. “It will run 24 hours.”

How it works

The infrared camera is mounted high up over an intersection, busy street or square – it requires such height so that its view isn’t obstructed by the many tall buildings in a city. The camera then collects the thermal images, which are analysed by computer algorithms – Gade’s speciality.

“It’s very hard, because the resolution of the camera is quite low,” she said. “It can be only a few pixels that we try to distinguish… We don’t always get a good profile of the person, so it can be hard to distinguish between a person, a small car, a cyclist or even a dog. It’s an issue we’re working on.”

The images are also overlaid onto a matrix to fix them in geographic space, so that each pixel in the image can be matched to a real-world co-ordinate. This is the work of Søren Zebitz Nielsen, a researcher at the University of Copenhagen. “You find out what real-world co-ordinate each pixel refers to, more or less, and you do that for eight or 20 pixels. Then you [can build] a matrix between the ground and the image,” he said.

Finding answers

All that data lets researchers see a pedestrian or cyclist move into the zone, and track how they use the space. Are cyclists able to flow through intersections, or do they sit dangerously amidst cars, waiting for their turn to cross? Do pedestrians linger in an area, or speed-walk through? The data collected can help answer such questions.

And this isn’t only theory: the system is being used in practice. It’s already been enlisted indoors at sports arenas to record fans’ movements, while in the Netherlands such cameras have tracked intersections where cars and cyclists are forced into close quarters. “They were trying to [capture] situations where cars and cyclists were close to hitting each other,” said Gade, referring to a specific crossroads where cars might turn into a lane of cyclists going straight. “These situations where they’re close to each other, we were looking at the flow in the car lane and the flow in the cycling lane.”

Sustrans now plans to take the work of the researchers and apply it here in the UK. Oram believes it will likely first be tested in Edinburgh – at a location handily near to Sustrans’ offices, where a complicated junction has a cycle lane running through it.

Building smarter cities

The project could give city planners the data to improve streets and communities. “The computer can carry out quite complex analyses of how people are behaving in a public space,” Gade added. “It can count the number of people who are lingering in a given space, and tell you how long they’re lingering for. This has implications for the economy of those places and how liveable they are,” she said.

For example, a camera could show that people are speeding through a particular square, and not stopping at shops or parks in the area. City planners could, say, add benches or trees, or make other minor changes, and then look at the camera data to view their effect – a “feedback loop”. If the results are positive, that particular area could be developed by adding more greenery or seating areas in order to make it a more attractive option for people to choose that route, or encourage them to linger longer.

“We’d have the evidence to show people are spending longer here,” Oram said, “which would be quite a good case for the changes.”

Indeed, while the project is focused on cyclists and pedestrians, Oram said that on a wider level it’s about how we think about urban design. “I think there’s a case for changing how we think about cities,” she said. “Instead of thinking of them as spaces and places and locations, [we could] be thinking of them as actions and interactions and transactions.”

Final thoughts: Smart tech gets the green light

A car that can change all red traffic lights to green? Now that is smart – and developers at Newcastle University are currently trialling such a technology in Newcastle city centre.

The device, which attaches to the windscreen of a vehicle, communicates with traffic lights to ensure a smooth, continuous journey. Some vehicles – such as those used by the NHS to transport patients – receive priority, while the rest of us are told to adjust our speed to ensure we see only green signals.

“For example, the system might advise a driver that if they travel at 24mph, they’ll hit the next four sets of traffic lights on green,” said Phil Blythe, professor of intelligent transport systems at Newcastle University. “In more congested areas, or at particularly busy times of the day, vehicles on key roads might be given priority in order to keep the traffic flowing.”

Aside from traffic signals, the system can warn drivers of an accident up ahead, or alert them if someone has jumped a red light at an upcoming intersection. So far, 20 sets of traffic lights and 14 vehicles have been fitted with the experimental technology, with more to follow.