The Good and the Bad of Driverless Cars for Cities

Driverless vehicles in a 2012 test in Spain. (Volvo)
Driverless vehicles in a 2012 test in Spain. (Volvo)

The age of driverless cars is rapidly approaching, and no one seems to know what to do about it. The technology is picking up steam in the behemoth automotive industry while only a few states have regulations on the books for autonomous vehicles (AVs). Just this week General Motors penned a half billion dollar investment in Lyft to develop AVs, and last month Google and Ford announced a similar partnership. The implications for the world’s transportation systems and urban living is uncertain, but there will surely be a mix of positive and negative impacts. This post explores some of the more likely ideas.

Before diving in, a quick primer on the five levels of automation defined by the National Highway Traffic Safety Administration:

  • No-Automation (Level 0): The driver is in complete control of the vehicle.
  • Function-specific Automation (Level 1): Automation of one or more specific control functions. Examples include electronic stability control or pre-charged brakes.
  • Combined Function Automation (Level 2): At least two control functions can be automated at the same time (e.g., cruise control in combination with lane centering).
  • Limited Self-Driving Automation (Level 3): The driver can cede full control of all critical functions under certain traffic and environmental conditions, but there must be adequate transition time for the driver to take over.
  • Full Self-Driving Automation (Level 4): The vehicle performs all safety-critical driving functions and monitors roadway conditions for an entire trip, including occupied and unoccupied vehicles.

There are two big issues that muddle the AV issue: adoption rate and energy sources.

The adoption rate is the change in proportion of AVs versus the human-driven vehicles of today. This will probably start out as a slow process, where most car manufactures offer an AV model or two at first (we can expect this by the early 2020s) and then gradually they will adopt it for most or all of their products. Within 10 or 20 years, when there is a critical mass of AVs on the road and the benefits of AVs are publicly accepted, governments will impose some type of restrictions on human-driven cars. If they’re not banned outright, in progressive areas drivers will certainly be required to have more stringent licensing and insurance coverage. Some regions may choose to restrict human-driven cars to rural areas or test tracks, or for special purposes like utilities, emergency services, and commercial deliveries in urban areas. Personally, I like driving on road trips and wouldn’t mind taking the wheel when heading to the mountains or coast for skiing or camping; plus, things like snow, mountains, and trees may inhibit lane sensors and GPS signals from reliably navigating AVs outside of the metropolitan areas.

Despite this relatively short timeline, U.S. cities are unprepared for the arrival of driverless cars. Local and state governments have adopted few plans and regulations that deal with AVs in the realms of liability, transportation planning, and engineering standards. This may be because AVs are still perceived as fringe technology that doesn’t merit much attention. But as AVs improve and become more available, regulators and city planners will need to pay attention and decide how they will accommodate a significant upheaval in how the vast majority of Americans get around.


The other issue is energy. Academic studies like this one for Lisbon, Portugal seem to indicate that AVs will make traveling by car so easy that vehicle-miles-traveled (VMT) will increase and low-density urban sprawl will propagate further. If AVs are primarily reliant upon fossil fuels, this will have devastating impacts on our environment. Carbon emissions will accelerate and energy companies will be more willing to extract oil from difficult and sensitive places like tar sands and under the oceans.

This scenario might be avoided if Elon Musk’s vision comes to pass. Musk, founder of California-based Tesla Motors, is certain most of the world’s vehicles will soon be electrically powered. There is reason to take Musk seriously. As detailed in a Wait But Why post, in 2003 Tesla entered the auto industry by storm as the first startup car company since the 1920s, and it has successfully produced popular high-end electric vehicles. Their business plan uses revenue from those sales to fund the development of cheaper vehicles that are affordable to the middle class (with the first such model available in 2017) and doubling the world’s lithium ion battery production with a massive battery factory now under construction in Nevada. Tesla’s Model S already has a range of 240-plus miles, more than enough for most trips, but to further ease range anxiety the company has deployed a nationwide network of superchargers that can recharge batteries in minutes.

Energy flows in the United States. (Lawrence Livermore National Laboratory)
Energy flows in the United States. (Lawrence Livermore National Laboratory)

Critics argue that electric cars have questionable environmental benefits because much of the electric grid in the U.S. and the world is powered by fossil fuels, namely coal and natural gas. But centralizing combustion in power plants, instead of individual cars, leads to lower overall emissions; the federal Environmental Protection Agency offers a national and state-by-state comparison of EV’s carbon impacts. Further, renewable energy technology prices are rapidly falling and more clean capacity is coming online every year, lowering the carbon footprint of electric cars over time whereas gasoline cars have a fixed carbon footprint.

There is reason to be optimistic about the energy picture. Solar panels, for instance, now cost just 74 cents per watt (compared to over $70 per watt in the 1970s), and installations are expected to grow. To illustrate the potential of solar, powering the entire U.S. would require only enough panels to cover 10,000 square miles (.003 percent of the contiguous U.S. land area). Of course, the U.S. already has a number of other renewable sources like nuclear, hydro, and wind that can be scaled up (and Congress recently extended the tax credit for wind and solar power). The Obama administration is also aggressively working to shut down hundreds of coal plants.


Anyway, back to driverless cars. We’ve established that even if AVs increase the amount of miles driven, their environmental impact will be less than today’s cars if they help lead the transition to electric vehicles and a renewable electric grid. There are two other primary benefits that driveless cars will have for cities: traffic safety and transportation efficiency.

The safety advantage will be massive. Some 32,000 Americans die each year in car collisions, about the same number killed by firearms. That’s also equivalent to a 9/11 attack every five weeks, but it goes virtually unnoticed in the national media. Globally, the World Health Organization estimates 1.25 million (that’s 1,250,000) people die each year in traffic crashes and millions more are injured. Half of those killed are vulnerable users like pedestrians, bicyclists, and motorcyclists, and motor vehicle crashes are the leading cause of death for 15-29 year olds. These numbers are staggering and horrific, and they alone are worth the pursuit of driverless cars and better engineering standards.

Features of Google's prototype driverless car; previously the company modified existing vehicles for testing. (Google)
Features of Google’s prototype driverless car; previously the company modified existing vehicles for testing. (Google)

Driverless vehicles will be much safer than human-driven vehicles. As any American commuter can attest, drivers are often in a rush and are prone to distraction, fatigue, poor sight, long reaction times, and road rage. All of that uncertainty is eliminated when a computer is in control. AVs use lasers, radar, and cameras to keep a 360 degree view in all directions at all times. These tools help anticipate the movement of other road users and detect problems far ahead. Being more cautious around vulnerable users and programming strict adherence to traffic laws, like stopping at red lights and obeying speed limits, will dramatically reduce traffic injuries and deaths.

As an example, Google’s prototype driverless vehicles have driven 1.3 million miles and experienced 17 minor collisions; Google’s November 2015 report (PDF) says none of them were the fault of the AVs. But the degree of safety improvements will depend on the rate of AV adoption, which will be influenced by prices and regulations.

One issue that needs to be worked out is the matter of liability in the case of a collision. If an AV vehicle, despite its capabilities, hits and kills a person walking, who is at fault? Is it the occupants or the vehicle’s owner? If the vehicle is empty, would it be on the AV manufacturer or, if a taxi, the service operator? Or will society continue to shrug its shoulders and say the victim should have worn brighter clothing? Whatever the case, accountability in the form of stiff financial penalties and jail times needs to be established now. Unclear penalties and lax enforcement will lag public support for AVs.

As for the second benefit, efficiency, advocates often illustrate a future where AVs travel in platoons or “road trains”, groupings of fast-moving AVs that coordinate their movements for higher utilization of roadway space and more predictable traffic movements. This concept hinges on a network of connected vehicles that can talk to each other and roadside sensors, potentially smoothing out congestion, improving safety, and lowering emissions.

There are three other dimensions to the efficiency discussion that naturally follow: the allocation of road space, parking requirements, and transportation funding.

Another benefit of driverless cars: they'll be a lot more patient. (Photo by the author)
Another benefit of driverless cars: they’ll be a lot more patient. (Photo by the author)

First, if drivers will no longer be behind the wheel they will likely be doing other things like work, watching movies, reading the news, etc. This means they will be less aware of traffic conditions, other vehicles, and stop lights, and therefore have less of a reason to be emotionally connected to traveling. This means there will be much less political opposition to more humanized use of urban streets, like converting general lanes and on-street parking into transit-only lanes, bike lanes, parklets, and wider sidewalks. It will also be easier to implement pedestrian-friendly tactics like curbless streets and longer crossing times. In short, driverless cars will give planners and engineers leverage to build more pleasant urban environments.

The second point, parking requirements, depends greatly on the ownership model of AVs: will people own AVs like they own cars today, or will people call for a car to pick them up and drop them off without the worry of parking, like a taxi service? This is a false bifurcation, and it’s likely that there will be a mix of both just like there is today. High-income, suburban, and rural residents will continue to have the means and practical needs to own their own vehicles. In urban areas where fewer people own cars and there is a greater population density,there will likely be a heavier mix of Uber-like services where AVs circulate on-demand. Cities could mandate that these automated taxis use centralized parking and fueling/charging facilities, enabling planners to lessen or eliminate off-street parking requirements for most new buildings. This would dramatically lower construction costs and therefore rents, making cities more affordable to people and businesses. It will also free up more land and financial capital for buildings rather than surface parking lots and standalone garages, enabling greater urban density.

The third point, transportation funding, is a result of the main argument behind AV efficiency. If traffic jams become a thing of the past and AVs can move faster and bunch together, then the world’s roadways will suddenly have an enormous surplus in capacity. In the U.S., AVs’ high fuel efficiency and likely adoption of alternative fuels will cause highway funding to decline further (assuming state governments and Congress continue to not raise gas taxes). The American interstate system’s many branches and spurs in urban areas will be excessive and an even bigger maintenance drain then they are today. This will give planners momentum to narrow and demolish unneeded freeways and suburban arterials, freeing up funding for more critical infrastructure like affordable and efficient mass transit.

Balancing Act

These last two points reveal a potential issue: commuters who ride their personal AVs from the suburbs to downtown, and then to avoid parking costs send their vehicle home or direct it to circulate city streets all day. This has negative implications for congestion and emissions. The best way to counteract this will be a relatively high road usage fee, something that state governments like Oregon and Washington are already considering to replace dwindling gas tax revenues (see my summary of Oregon’s pilot program). Cites will also need to implement congestion charging in their densest areas, like London and Singapore do, to further discourage empty vehicles from roaming city streets and wasting fuel.

In the shared-vehicle model, if the average Lyft or Uber fare is any indication, low-income and even middle-income people will likely be priced out for frequent daily trips. But, on the flipside, AV technology will inevitably find its way into the world’s transit systems. This is already a reality on a number of rail systems (like in Vancouver, B.C. and at airports around the world), but the largest impact will be felt with the workhorses of major cities: local bus routes.

Hourly costs for operating a King County Metro bus route. (King County Metro)
Hourly costs for operating a King County Metro bus route. (King County Metro)

Once transit agencies realize the enormous cost savings they will reap by replacing bus drivers with computers, the transition will be relatively quick. Wages and benefits for King County Metro, for instance, take up nearly half of the operating budget (PDF). Extra space in the budget will allow for enormous improvements in service frequency, passenger amenities, and system capacity, making transit much more affordable and accessible to riders of all types, especially people who can’t afford or don’t want to use cars. To be sure, human employees will still be needed for planning and maintenance. And routes that serve places like hospitals, immigrant communities, and senior centers may benefit from a human touch. But overall, on most routes there is no reason buses couldn’t be fully automated like other passenger vehicles within the next decade or two.

This hints at another socioeconomic issue: job losses caused by automation. Many people have jobs driving buses, taxis, and trucks; in the U.S. alone, 4.6 million people are employed in the broadly defined transportation sector. It’s a daunting prospect, but not unlike earlier transitions following the inventions of steam power and computers. The transition will probably be gradual enough to avoid mass unemployment and create many new job opportunities in emerging fields.


Autonomous vehicles will have major impacts on energy consumption, traffic safety, transportation efficiency, and socioeconomic equity. The degree of these effects are largely up to the adoption rate and government regulations, but it’s indisputable that driverless cars will be on the road in large numbers within five years. City planners and elected leaders need to leverage this change now rather than reacting to it later, making improvements to land use planning, street designs, and mass transit that will benefit the greater good.

This article is a cross-post from The Northwest Urbanist.

AV, I’m Walking Here! Pedestrians and Self-Driving Cars

We hope you loved this article. If so, please consider subscribing or donating. The Urbanist is a 501(c)(4) nonprofit that depends on donations from readers like you.

Scott Bonjukian is a car-free urban designer with a passion for sustainable and efficient cities. With degrees in architecture and urban planning, his many interests include neighborhood design, public space and street design, transit systems, pedestrian and bicycle planning, local politics, and natural resource protection. He primarily cross-posts from his blog at The Northwest Urbanist and advocates for a variety of progressive land use and transportation solutions.

Inline Feedbacks
View all comments

Good stuff. I agree with most of the ideas. I disagree with one thing, though, and that is the idea that it will decrease congestion. There will likely be a very long period where many of the cars are not automated. During this period, automated cars will like increase congestion. Automated cars tend to be very cautious (this is one of the problems with the Google cars). Human drivers are way more likely to risk an accident than an automated car (which says something about society, but no need to go there). So, for example, a human driver is willing to tailgate in heavy traffic, but an automated car won’t. An automated car will slow down to the speed that is posted on the freeway signs (e. g. on I-90). This means that an automated car will slow down to 35 MPH, when everyone else is willing to go 60. Of course the person driving 60 will eventually have to slow down, but in the mean time, they pass the automated car. So while everyone would be better off if everyone had an automated car, a world of half automated, half manual cars means just as much congestion, but slower overall speeds for the automated car.

This is very important, because it feeds into the other conclusions. Without a doubt the suburban commute looks a lot better when you can just sit back and do your business. But the folks who struggle with traffic every day do so mainly for one reason: it saves time. Very few work downtown. They work in places like First Hill, Fremont, Ballard, Georgetown or Queen Anne. Taking a couple buses is just too slow. Better to slog through the traffic (as frustrating as that is) than to enjoy quiet time on the bus (or train). With transit you can get work done, you can read, you can relax. But when you drive, you get to see your family sooner. To many (if not most) that is more important, which is why fast transit is way more important than comfortable transit. It is also why fast transit will still be favored over slow cars, even if the slow car is suddenly a lot more enjoyable (because you can read, etc.).

So the idea that automated cars will usher in a new era of suburban living seems unlikely. After all, it isn’t just the commute that is keeping people in the city, it is the fact that people would rather live there. But the bigger issue is that cars don’t scale. Nor will they ever scale. They just take up too much space. Even if the automated cars were perfect, even if 100% of the cars were running down the road at 60 MPH just inches away from each other (because the software is perfect) you still take up way more room than a set of buses carrying the same number of people.

Which is not to say that automated cars can’t or won’t cause urban congestion. I think this is likely. Remove the parking issue, and lot’s of people will drive downtown in the middle of the day. Your car becomes a cheap taxi. But then again, remove the driver, and the taxi becomes a cheap taxi. Much cheaper than your car, since the use of it is shared for many more trips.

Predicting the future is very difficult. It is quite possible that a fully automated car, like atomic fusion, is always a few years away. But my guess is that automation, like the new taxi companies (Uber, Lift, etc.), will lead to less car ownership, not more. People want to live in the city, they prefer transit (when it is faster) and would rather not have to deal with owning a car, unless they use it a lot to get out to the countryside. Even that is a possibility — an automated taxi to Mount Rainier? Sure, why not.

Bryan Kirschner

You’re not only right that automation will lead to less car ownership rather than more, but also all the automakers know it and are telegraphing the fact that they see the writing on the wall: GM’s investment in Lyft, BWM’s DriveNow, etc. There is massive deadweight loss in the individual ownership scheme and they all know it, so they are re-positioning to become more like fleet management / shared vehicle services.

(Edited to add: for the geeks out there, this is parallel to data center server utilization versus cloud computing on demand. Cloud wins. Not even close.)

Doug T

I agree with Ross that most of the city-wide benefits of AV don’t accrue until the whole fleet is automated. A few motorists can gum up the works pretty easily. It’s going to take passing a law and enforcing it heavily to finally get all the motorists off the streets in cities to usher in the pedestrian paradise. Until then removing curbs, parking and dangerous motorist behavior is going to be touch and go.

Still, AV presents some exciting possibilities. Good article, Scott. I hope we can steer AV technology and laws toward the most benefit to cities.

Ross, as a side note, Ross, is that true about most car trips being towards Seattle neighborhoods rather than the CBB? I had been under the impression the central core was drawing most of the car trips and congestion.


I suppose it depends on what you call core. Is South Lake Union core? If so, then how about lower Queen Anne (which really isn’t that far away)? What about First Hill?

Anyway, my understanding is that a very low percentage of the people who go downtown go there by car. It is a pretty common refrain. If you want to go downtown, take a bus; if you want to go somewhere else, drive. The main reason is that it is usually faster to take a bus downtown, while it is usually faster if you are headed somewhere else. I have no study to point to, but the vast majority of people I’ve talked to say speed is the most important reason they take or don’t take transit.