AI-based traffic control gets the green light
At the end of my Melbourne street there’s a new system being installed for traffic management. I hadn’t even noticed the extra cameras, vehicle and pedestrian sensors, LiDAR and radar on the intersection, but these tools are all part of a larger system, with researchers hoping that a 2.5km section of Nicholson Street, in Carlton, will eventually be run by an artificial intelligence (AI).
This might sound a little nerve-wracking to the average commuter, but these “smart corridors” are popping up around the world – systems that promise to provide us with less traffic and better safety.
“Many cities around the world have dedicated corridors or smart motorways that are equipped with sensors, CCTV cameras and AI for predicting the traffic flow, speed, or occupancy at a specific moment in time,” says Dr Adriana-Simona Mihaita, an AI infrastructure researcher at the University of Technology Sydney, who was not involved in the research.
“Accurate predictions will provide transport operators with the means to make informed decisions and apply new control plans, or adjust the current ones according to ongoing traffic or eventual disruptions.”
Even without AI, our current traffic light systems are complex technology. Detectors under the road surface clock the presence of vehicles and determine whether the lights change, and how long the green lasts for. The “push button” changes the green walk display, and some detectors even determine how fast or slow the pedestrians are moving. This is all controlled – in Australia at least – by a system called SCATS, or Sydney Coordinated Adaptive Traffic System.
This is an “intelligent transport system”, but it’s not machine learning or AI. Think of it as a calculator, but the values being input are cars and pedestrians, not numbers. And SCATS does have its flaws. To start with, it’s unable to see cars coming – only registering them once they’ve arrived on the detector. And the system is also not particularly good at including other modes of transport such as trams, cyclists and pedestrians.
This is where Nicholson Street – home to plenty of cyclists and the 96 and 86 trams – will come in handy.
“With SCATS there are loop detectors that tell us how many cars are in the queue from all directions, but we don’t really see the number of cyclists, or pedestrians, and we don’t take their delays into account,” explains one of the researchers on this Nicholson Street project, University of Melbourne transport engineer Dr Neema Nassir.
“As long as we build our environment around cars, and prioritise their right of way over other modes of transport, we are promoting people using that mode of transport.”
Using 180-degree high-definition cameras, as well as a range of detectors (including the normal SCATS detectors), Nassir and the team of researchers are currently testing the AI system using this real-world data in a computer simulation.
When the AI eventually starts testing and directing traffic in the real world, it will be done using “edge computing”. This means that the AI-based traffic optimisation will happen at the intersection or “node” rather than at a central system. With the sensors taking and analysing the data almost instantly, the lights might change if there are more pedestrians waiting, or a tram might get right of way if it’s running behind schedule.
Nassir hopes the intersections will be safer, trams will run more evenly, and there will be less stopping for cars on the road.
However, there might also be some hiccups on the way.
Safety is the top priority for the system, with efficiency coming in second. This means that the AI will be more likely to cause traffic jams than accidents.
“If everything goes wrong with the algorithm and with the computations, it’s more likely that we may end up with a gridlock as opposed to safety concerns,” Nassir says. “We’re talking about an intersection that is designed to be robust enough that it can operate even when the traffic lights are off.”
The AI will be more likely to cause traffic jams than accidents.
Having humans in this scenario, who are able to stop if required, is actually helpful. Unlike an autonomous car, which needs to function in an almost unlimited number of circumstances without a human behind the wheel, traffic systems are comparatively simple. And, if something was to go wrong, humans are able to make a judgement and stop or swerve if required. Plus, because SCATS is already automated, it means that that the commuters coming through Nicholson Street might not even notice the change.
But that presents another dilemma – is it okay to record all this extra information and send it through an AI to make decisions?
“The most important type of sensors are high-definition cameras,” says Nassir. “These are mounted high on poles and have 180 degrees of coverage, up to 50 metres down each approach. These are coupled with image-processing software that can help us detect and register and classify different types of passengers.
“We also work with the data from key cards on public transport . There are tight regulations and rules regarding this personal data. It is always anonymised and protected.”
Because SCATS is already automated, the commuters might not even notice the change.
Nassir says the cameras are not capable of being used for facial recognition. In a world where facial recognition is happening every time you open your phone and in stores like Bunnings and Kmart, traffic lights are probably not the place we need to be too concerned about our privacy being invaded.
“Several phone applications that are currently in use today are already collecting private mobility data, together with personal preferences and route choice patterns, which represent a deeper concern for daily transport choices,” says Mihaita.
“Similarly, public parking areas in large shopping malls have automatic plate recognition capability and store daily information on all vehicles entering/exiting the malls, which could be seen as personal information shared with the consent given while entering the parking area.”
But ethical issues don’t just stop at these records. According to Professor Toby Walsh, an AI researcher from the University of New South Wales, there’s a number of ethical questions we should be aware of as these systems become integrated into our daily lives.
For example, if our traffic systems know who we are, it might not just be a case of prioritising cars over other forms of transport, but instead the rich over the poor, or the paying verses the non-paying.
“At stake are fundamental issues of fairness and justice,” Walsh explains. “You might start having to trade off my journey time against your journey time. Who gets priority?
“Then there’s an environmental ethical issue: are we encouraging people to make more individual car journeys by improving traffic flow? Should we actually be trying to discourage people from getting in cars, and encouraging them to Zoom for work or get public transport?”
Although the Nicholson Street AI project is trying to balance the priorities of trams, pedestrians, cyclists and cars, easing urban congestion is also an important part of the project, and as Walsh says, “Traffic is like an ideal gas that expands to fill the roads available.”
“You might start having to trade off my journey time against your journey time. Who gets priority?”
Professor Toby Walsh, UNSW
Despite these questions, Walsh argues that even if AI isn’t perfect, humans are worse. He has been involved in the research for another AI intersection – a particularly busy roundabout in the south-west of Sydney.
“A thousand people are going to die in Australia in the next year, caused by traffic accidents. Almost all of those accidents are caused by human stupidity. Almost all of those accidents wouldn’t happen if we ceded our human control and all of our misjudgements – all of our texting and drinking and driving – to machines,” he says.
“There’s always going to be unintended consequences – random shit happens, and the death rates are never going to be zero. But it would be a small fraction of what it is today.”
This article was originally published on Cosmos Magazine and was written by Jacinta Bowler. Jacinta Bowler is a freelance science journalist who has written about far-flung exoplanets, terrifying superbugs and everything in between. They have written articles for ABC, SBS, ScienceAlert and Pedestrian, and are a regular contributor for kids magazines Double Helix and KIT.
Image: Getty Images