Q&A: A Traffic Engineer on the Future of Autonomous Vehicles

Autonomous vehicles are the future. They’ve quickly moved from an experimental technology to having a foreseeable place in society’s future — in no small part thanks to companies like Google that have made testing of driverless cars a big priority.

Kinks abound, and it will be years before we see people in the (former) driver’s seat with their feet on the dashboard taking a nap, but it’s already worth considering how roads, traffic practices, and related infrastructure will change in the future, as autonomous vehicles begin to roll out. Inverse spoke to Michael Palamara, a transportation engineer working in Sydney, for some insight as to how we might expect our roads to transform.

To what extent are places like Australia planning for driverless/autonomous cars?

At the moment there aren’t any major infrastructure projects tied to autonomous cars. They’re currently testing the vehicles in current conditions and testing out specific routes. So they’re making autonomous cars work with the standards we’ve got now, like roads with turning circles.

The things that will change are when there is a transition period, which is probably going to be one of the most dangerous periods, which might start when companies or governments say they’re going to start introducing autonomous cars now for the next 10 years. Right now there isn’t anything planned, however.

What are the major driving practices on the road that would be expected to change because of autonomous vehicles?

I think human behavior will be one of the biggest factors. For the Google [autonomous] cars, the majority of the accidents they have been involved in were rear-end collisions caused by other, human drivers.

Cars will need to programmed specifically for the roads. At a stop light, when you see a green light, you wait one of two seconds or however long to make sure it’s clear before you car. You’ll have drivers interacting with driverless cars, and this might cause confusion and problems, since you don’t have another human face to communicate to.

But if you remove human error or human behavior, you’re removing uncertainty. The road environment is full of uncertainties, like other drivers, pedestrians, animals, and so much more. As a driver, you learn how to automatically react to those things, like slowing down when you see a child close to the road. Driverless cars remove many uncertainties, but others who are working or playing or walking on the roads will need to learn how to adapt to technology that might not otherwise react as quickly.

How will infrastructure need to adapt? Are there certain road structures or layouts that driverless cars work best with? Others that would be expected to be demolished or converted into something else? Let’s assume it’s at a point where 90 to 100 percent of all cars on the road are autonomous.

If those vehicles are able to communicate with each other — and this gets pretty far down the timeline — and you’ve got a road network they’re connected to, those vehicles can move together, and sort of act as one. So if you merge onto the highway, and speed up and slow down accordingly. And as you enter the highway you’ve got to adjust to get in. So think of cars on the highway that are a mile back. By the time the car gets on the offramp, those cars a mile back know there’s a car going to merge, and they slow down appropriately or move into another lane.

You may see there may not be any traffic lights, because if you’re in a system where each car knows where the other cars are, they’ll adjust their speed accordingly. The easiest example is on a highway, where one lane has a capacity to hold a certain amount of cars. But this depends on speed, where you need more distance between cars so people have enough time to react when something surprising happens. On a highway with automated cars, the road environment may become more enclosed, since cars can travel closer together. You can have an efficiency in the network that reduces travel time or reduces congestion. It’s all self-managed. People from side streets or non-prioritized roads may also get in more quickly to major roads.

What about parking? With automated parking maneuvers, would we perhaps expect to see changes in parking layouts? Like less lots and more parallel spots?

Parking also comes down to the kind of communication these cars will have with one another, since effectively any space you have could now act as a parking spot. Cars could just stack in next to each other. If you have to get to your car it might know that these four cars next to it will have to pull out, which could work out.

Still, parking wouldn’t necessarily change very much. Cars are around five meters in length, so theoretically in a 10-meter space, you could park two cars. But in order to actually park the cars, you would need something like 11 to 13 meters in space. Efficiency would increase based on being able to park cars next to each other, and having them move in and out as needed, without the driver necessarily being there.

That depends, though, on how autonomous cars change ownership of vehicles. We might see less and less privately owned cars, and a car simply arrives at your house or wherever you are based on when you need them. But that’s a separate issue.

With more cars on the road because of autonomous driving, could we also expect urban areas to create more pedestrian-friendly walkways or bikeways?

It definitely would. If you remove the need for everyone to own a car — since everyone now has access to a car — then you don’t need as much road space. You can also increase efficiency of the road network through automated vehicles. Instead of three or four lanes for traffic, you might now only need one or two for an optimal driving environment. There might not be on-street parking, just parking bays for unused cars.

You won’t have people randomly stopping, or turning left when you’re not supposed to, or having cars block the intersection. There are all of these tiny little things that happen every day that add up and create a lot of wasted time and space. In cities, a minor incident could back traffic up three or four blocks down the road very quickly. Even small efficiency changes would give more space to people walking, while traffic still flows.

Are there any other effects the mass adoption of driverless cars would have on road infrastructure?

Things go back to what the transition period would look like. You’d probably see a time when driverless cars are first getting onto the road network, and the government starts to optimize the infrastructure. It wouldn’t be immediate — it would probably take at least five or 10 years to change everything in the system. The infrastructure will change, but it will be a reactive effect — not a proactive one.