How Growth in Autonomous Vehicles Could Improve Traffic Flow

By Graham Jarvis

---

16th November 2022

Congestion is a growing problem and so a number of solutions have been proposed.

They range from encouraging people to leave their cars at home, discouraging private vehicle ownership, ride sharing and by promoting other forms of transport through shared mobility schemes. Now, in Oxford and Canterbury in the UK, they are proposing to fine people for travelling out of their designated districted a certain number of times a year. The latter, which some commentators argue designed to limit freedom of movement in private vehicles, brings forth a sense of Big Brother in George Orwell’s novel 1984.

Autonomous vehicles (AVs) will be part of the shared mobility paradigm, although there may still be some private ownership of them. In the future most autonomous vehicles will be called up via an app rather than be waiting on someone’s drive ready for them to use. The aim will be to reduce the number of vehicles our highways, thus cutting traffic congestion and pollution levels.

However, this won’t occur overnight and many people will most probably still want to own their own car in the future. Even Carnegie Mellon University recognizes that private ownership won’t be a thing of the past for some years to come – if ever. Yet, its study has found that if at least “20% of cars are autonomous vehicles, traffic systems may start to see the operational improvements these vehicles are expected to bring,” writes Skip Descant for Governing.com.

Vehicle collaboration

Carlee Joe-Wong, associate professor in the electrical and computer engineering department at Carnegie Mellon University, and one of the authors of the report, Mixed-Autonomy Era of Transportation: Resilience and Autonomous Fleet Management, explains the reasoning behind its findings:

“We felt autonomous vehicles would be better able to collaborate with each other. Autonomous vehicles are more likely to have an internet connection that allows them to communicate, and they have onboard computers that permit them to search for different actions that serve to find the best outcome.

“Our thought, given these capabilities, is that autonomous vehicles can take actions to benefit autonomous and human-driven vehicles on the road. I think it’s possible to have both on the road as not all vehicles will become autonomous all at once. It will take time for the transition to happen. We are just starting to see the initial autonomous vehicle deployments, with human safety drivers. So, I guess 2050 will be the earliest when we will see them dominating the road.”

Modelling assumptions

Most of the conclusions in the policy report are based on modelling assumptions because there isn’t much data available at the moment to confirm what would actually happen in reality. Much, she explains, depends on how autonomous vehicles will be deployed. “If you have fleets of AV taxis or delivery vehicles, they will be highly co-operative but individual vehicles may not be so much – depending on the regulations governments put in place,” she comments. There may also need to be some incentivization to increase the usage of autonomous vehicles such as reduced tolls, or reduced parking fees to encourage the vehicles to collaborate.

Commenting on the report’s findings, she adds: “For the 20% threshold of autonomous vehicles that we found, we were looking at a simplified map of Pittsburgh and we were using data on typical traffic flows. Using those data, we were able to simulate the behavior of autonomous vehicles, showing there would be improvements in traffic time.”

As a result of the modelling, despite admitting that some assumptions were made which may or may not be accurate, the authors of the report concluded that autonomous vehicles will be co-operative. Yet, there could be software bugs, or other factors that could prove problematic.

Time will tell how co-operative they will be. However, they may be more co-operative because of their lack of human temperament. As autonomous vehicles will account for 1%-5% on the road by 2025, it will be a long while before the reality of their impact on congestion is realized – particularly as the dates for seeing the widespread adoption of AVs is constantly changing – being pushed out.

Sharing information

Niels Peter Skov Andersen, chair of ETSI Intelligent Transport Systems Technical Committee at the European Telecommunications Standards Institute (ETSI) says an autonomous vehicle on its own can’t improve traffic congestion. However, the shockwave effect that creates traffic queues can be eliminated or reduced with vehicle-to-everything (V2X) communication.

He explains: “An autonomous vehicle without communication support will probably slow traffic down. With the communication component, you will be able to know that the car in front of you is going to brake. This would mean you don’t need to brake harder. So, it’s a question of sharing information of the intention of the car in front of you. This is what we think will help. Today, with Phase I we see it works and from a technical perspective we are planning for this for the subsequent phase towards automated driving. Technically we have started to get ahead of legislative framework. We need for the infrastructure to talk to the car.”

Legal framework required

To achieve this there is a need for a legal framework to give a legal status to the information that a vehicle receives from the infrastructure. This may be a speed limit that’s sent from a road sign. He believes it would otherwise be “difficult to put in place the technical capabilities and to standardize them.”

He adds that improved traffic flow will come from AVs that obey the rules of the road, allowing a road authority to optimize traffic flow. With better communication between the vehicles and infrastructure, it should also be possible to reduce the safe distance between the vehicles because true communication can reduce the required reaction time to a vehicle or obstacle in front of them.

Previous studies

Allanté Whitmore, director of the Autonomous Vehicle Initiative at SAFE, says she is excited by the findings of Carnegie Mellon University: “Prior studies estimated an AV penetration rate closer to 60%, so this is really promising and indicates progress. Again, the Carnegie Mellon University report arrives at the 20% estimate under a very specific set of circumstances.”

Those circumstances include organizing and guiding the AVs using a shared algorithm with the goal of maximizing social welfare in a mixed fleet traffic. They also need to account for “the self-centered behavior of human drivers,” and it depends on the vehicles having reliable data for autonomous vehicles to provide details of traffic patterns, data, and up to date information about any emergencies that are occurring at any point in a journey, or whenever current traffic event occur. She agrees that these conditions are feasible, but underlines that need for policymakers and automakers to act now.

She adds: “AVs can improve traffic flow through coordinated fleet movement within a transportation network. It is promising to see that lower levels of penetration will yield positive results because, unfortunately, the current regulatory landscape is lagging when it comes to AV deployment.”

In her view it’s a case of the more autonomous vehicles on the road the better because it’s argued that AVs drive more precisely and efficiently, leading to safer and less congested roads. There is nevertheless a gap at the moment between “the technical availability and the political infrastructure or levers to meet the large-scale AV deployment and connectivity we need to tip the scales,” she explains. Meanwhile, there are already several successful pilot projects and other tests taking place in both the public and private sectors that could make co-operative driving a connected reality soon.