Smart city connectivity essential to avoid AV jam

AV promise of improved mobility only happens when traffic signals talk to vehicles, otherwise self-driving cars could struggle at junctions.
21 August 2023

AV jam: self-driving cars could be too cautious without connected traffic lights, find researchers in latest study.

Getting your Trinity Audio player ready...

Governments and automakers are on a mission to encourage drivers to buy battery-powered vehicles. And while electric powertrains will make city streets sweeter on the lungs, simply replacing fossil-fuelled petrol and diesel engines with batteries and motors will do nothing for traffic congestion on the road. In fact, autonomous vehicle (AV) designs may even make journey times worse – based on a study published this month in the journal Transportation Research Record and available as a preprint on arXiv.

So, should car buyers avoid advanced driver-assistance system (ADAS) upgrades on their next vehicle order? It depends. AV models, and the features that step toward fully driverless cars, have dual benefits – improved safety for road users and more efficient journeys.

On the first count, vehicles bristling with a variety of automotive sensors – such as cameras found on Tesla’s popular EVs and automotive lidar and radar configurations adopted by other OEMs – can lookout for approaching incidents that human drivers may have missed.

Even relatively low-tech ultrasonic parking sensors, which have a short range and are most effective at low speeds, have nonetheless proven to be useful in preventing countless dents and collisions with pedestrians.

All things considered, it’s reasonable to view increased automation and sensor fusion that stacks multiple data feeds as a boon for safety. For example, automotive radar can bounce underneath the car in front to see obstacles that would be blocked from a human driver’s view.

How to avoid an AV jam

But answering the question of whether higher levels of ADAS leading to full AV operation will improve journey times requires additional information. And that detail turns out to be connectivity – as the researchers in the study, which focused on saturation headway and capacity at signalized intersections, discovered.

When vehicles and traffic control systems are connected, more road traffic can pass through junctions. So-called signal phase and timing (SPaT) data broadcast from traffic lights at intersections gives both human-driven vehicles and AV types the opportunity to adjust their speed to maximize road capacity.

But in the absence of these broadcasts and without supporting vehicle connectivity, even AVs with full self-driving capabilities will fall short of their mobility promises. “This is because those AVs are programmed to drive conservatively in order to reduce the risk of collisions,” explains Ali Hajbabaie of North Carolina State University, who took part in the study looking at the traffic dynamics of having different portions of connected, automated, and connected automated vehicles at a computer-simulated intersection.

“There is a lot of research showing that automated vehicles can improve safety. But our research here – which relies on computational modeling – suggests that if we want to also improve travel time, an increase in automated vehicles isn’t enough,” he adds. “We need vehicles that are capable of communicating with each other and with the traffic-control systems that manage traffic flow at intersections.”

The finding has implications for city planners hoping to improve mobility around town by giving AV operators the go-ahead to provide self-driving taxi services. This month, Waymo – which was established under Alphabet and can be traced back to Google’s self-driving car trials beginning in 2009 – received its driverless deployment permit from the California Public Utilities Commission (CPUC).

And the vote of confidence from the CPUC paves the way for Waymo to now offer a paid, fully autonomous ride-hailing service in San Francisco, US. But without smart city infrastructure, some of those AV benefits could be left on the table if developers find themselves having to dial down performance to give human road users more reaction time to changes in the driving environment.

However, information sharing between vehicles (V2V) and between vehicles and road infrastructure (V2I) could reduce the size of the time buffer that needs to be programmed into mobility algorithms. And, in principle, AVs could drive much closer to their potential limits with direction changes being telegraphed to surrounding road traffic to help the decision-making of connected vehicles following.

Other studies have shown how connected traffic lights can smooth the flow of vehicles through city roads. But what makes this study different is the investigation of how a mix of connected and unconnected road users impacts journey times and lane use efficiency.

The worst-case scenario, based on the group’s results, is a stream of 100% AV traffic with 0% information sharing, which reduces lane capacity by 20%. And the result highlights the productivity risk of having AVs circulating on city streets without smart city connectivity.

Certainly, urban planners who are struggling to convince budget holders of the need for greater V2I connectivity can use the data to support their case. And on TechHQ we’ve written about how smart traffic lights are improving journey times on public transport, as buses can be prioritized to help services stick to their timetables.

Developing vehicle autonomy hand in hand with smart transport infrastructure is a winning strategy and plays into the performance gains of technology such as 5G that supports low-latency, fast-acting communications.

Automation-ready transport planning

From May 2017 to April 2020, a European project dubbed CoEXist examined what’s needed for cities to prepare for increasing numbers of AVs driving alongside conventional vehicles. And this included understanding the differences between how human drivers and AVs approach unsignalled infrastructure such as roundabouts.

The cities chosen for CoEXist pilots were Milton Keynes in the UK, Stuttgart in Germany, Gothenburg in Sweden, and Helmond in the Netherlands. And the talking points raised are informative. For example, given the number of shared spaces in cities, how can operators prevent the passive nature of CAVs from causing delays as they are obliged to give way to pedestrians? And what’s the best way of integrating ride-sharing CAVs into public transport systems?

AV designs causing city streets to grind to a halt is definitely not part of the industry picture, but AV jams could become a real problem if the connectivity of other road users and infrastructure can’t keep up with the pace of self-driving progress.