A Self-Driving Future

At the intersection of driverless cars and car sharing.

Cars that drive themselves seemed like science fiction just a few years ago, but recent demonstration projects have shown that the technology is already here. Self-driving car technology, pioneered by Google, has advanced so quickly that its ubiquitous presence on city streets is now simply a matter of time. Boosters say that mass-market autonomous cars are only 3 to 5 years away; others estimate at least 10 years. No one doubts they are coming.

But ideas about how these cars will affect cities and the environment seem to be stuck in the past. People think of self-driving, or driverless, technology as something added on to personal cars. Personal cars, however, spend 95 percent of their time parked, going nowhere, and waiting until they are needed. It’s more likely that city dwellers will view this technology as a service, like calling for a taxi. In principle, an on-demand car service could offer the door-to-door mobility of car travel without the fixed costs and hassles of owning a car.

Self-driving technology could cause a societal shift away from private car ownership and toward vehicle sharing. Widespread use of on-demand car services would result in fewer vehicles, which would then transform urban land use. Cities could reap significant benefits: fewer parking lots, more relatively dense, walkable neighborhoods, cleaner air, and reduced carbon emissions.

Car-sharing programs like car2go offer spontaneous, one-way, pay-by-the-minute rentals within a prescribed service area. Members use a smartphone app to locate a nearby car, get to it, swipe a card to unlock, and drive to their destination. When they get there, users park the car at the curb and walk away, leaving it available for the next user.

Now imagine a mash-up of this popular model and Google’s self-driving car technology. The car-sharing fleet could be retrofitted with self-driving navigation systems. (Let’s call the hypothetical startup company “Car2Google.” Of course, other car-sharing services like Zipcar or even traditional rental car companies could jump into the game.) Layering self-driving technology onto this system would allow people to order a car from a fleet and have the car pick them up. It’s a taxi service without the drivers. Users would summon a car with their phone and wait comfortably indoors. The car would call or text them as it approaches. Users would then hop in, talk on the phone, or nap while the car drives to their destination. Once there, they can just walk away. The service charges their credit card an amount based on the time or length of the trip.

In addition to improving car sharing for the 1.7 million people who already use such a service, driverless technology would expand the appeal to a wide range of people, including those who:

  • are too young to drive;
  • are too old or infirm to drive;
  • have had their license suspended;
  • are too drunk or stoned to drive safely.

Parents could safely deliver their children to or from soccer practice or play dates simply by ordering a vehicle for them. Elderly people who should no longer be behind the wheel and disabled people who cannot drive would still have mobility, with the travel arrangements made independently or by caregivers. And everyone would benefit from impaired late-night revelers having a safe way to get home.

Does this sound like a fantasy? Rapid changes just in the last year bring this idea a lot closer to reality.

Self-driving vehicles would vastly increase the efficiency of car sharing. When not being used, cars would drive toward potential users instead of sitting idle and waiting for users to come to them. Instead of having workers going out into the field to service the fleet, cars could go to a central location when on-board sensors or user feedback recommend maintenance. A vehicle would take itself offline, drive itself to be washed or have its oil changed, and then put itself back online when ready to pick up another customer.

Driverless technology would also allow for a greener fleet. Urban trips are usually short hops ideally suited to electric vehicles. And when the battery level drops below a certain threshold, an autonomous vehicle could go offline and drive itself to the nearest available inductive charging pad to recharge.

Many households are already giving up their second vehicle—or sometimes even their only one–in favor of a mix of car sharing, bicycling, and using mass transit. A convenient, self-driving shared vehicle service could accelerate this trend. Traditional vehicles would likely still be needed for camping trips or to pick up a load of gravel, for instance, but these specialty vehicles could also be rented, perhaps from the same company.

This scenario of a city buzzing with fleets of self-driving vehicles may sound futuristic, but existing technology can already do everything described here. Automated car technology can also be applied gradually, one element at a time. Commercial airliners and many transit systems are already nearly fully automated.

The legal system and insurance interests are predictably wary of self-driving cars, and it will take time for individuals to grow comfortable with the idea of taking their hands off the wheel and their feet off the pedal. But legalization has already begun in several states, the Internet is rife with remarkable self-driving car videos, surveys show popular acceptance of the idea, and Google’s test models have a spotless safety record thus far.

A Google rendering of what a driverless car "sees."

A Google rendering of what a self-driving car “sees.”

Although the future is murky and the impacts of this system are unclear, if such a service took off, cities could enjoy a wide array of benefits.

Less traffic. Studies have shown that each car-sharing vehicle takes between 9 and 13 privately owned vehicles off the road. When people pay for driving by the trip, they see the real costs; therefore, carsharing members drive 44 percent less than personal vehicle owners do. Cruising to find parking spots would no longer be an issue, thereby eliminating a third of traffic in some urban areas.

Fewer parking spaces. More than a third of land area in some US cities is dedicated to car storage. If people shift away from personal cars and toward shared vehicles, most of those parking spaces would become unnecessary. Acres of empty parking lots and driveways in desirable locations would open up for more productive uses like housing, playgrounds, or parks. If the change happens quickly, urban land prices could tumble.

Safer streets. People are terrible drivers. Human error causes or contributes to more than 90 percent of car crashes. Getting behind the wheel is the riskiest thing most people do in a day. Self-driving cars would be programmed to never speed, make sudden lane changes, or let drunk or distracted drivers make life-or-death decisions. The lead engineer on Google’s project expects driverless cars to sharply reduce crashes of all kinds.

Cost savings. Personal car ownership costs about $9,000 per year. Among young people especially, rates of driving and car ownership are already falling. A convenient car-sharing service that saves money could convince people to sell their cars while allowing them to maintain their same level of mobility.

Environmental benefits. More efficient travel patterns, fewer vehicles, cleaner cars, and less driving overall would lead to better urban air quality and lower carbon emissions. Less land would be paved over for parking, reducing stormwater runoff and heat island effects. These sustainability benefits would be market-driven and would not require government subsidy.

As with any potentially disruptive technology, the benefits might not materialize in the way we predict, and the future might not be as rosy as it seems. Consider the techno-pessimistic future:

More traffic. Economists are quick to point out that people usually consume more of something when it becomes cheaper. If people can sleep or work on their laptops while in a car, perhaps they will drive more and commute even longer distances. We simply don’t know how much unmet, latent demand there is for vehicle travel. In addition, trips that previously did not exist would result from new users (children, blind people), people who shift from mass transit, and the self-redistribution of empty vehicles driving to the next user.

Sprawl. Currently, only a small number of workers are super commuters who live an hour or more by car from their jobs. But if technology can eliminate the fatigue and aggravation of driving, more people may choose this lifestyle. The boom of streetcar suburbs in the 1890s and, later, auto-oriented suburbs in the 1940s showed that new modes of transportation have big effects on where people choose to live.

Less safe streets. If fewer cars sit parked and more of them are constantly circulating, streets could become busier. Early testing has been very promising, but self-driving cars have not been set loose on the full range of chaotic real-world conditions such as erratic drivers, snow-covered streets, and software or mechanical glitches.

Job losses. Certain workers lose out in this imagined future. Cab drivers, parking lot attendants, transit workers (who will likely be displaced not only by driverless cars but also driverless buses and trains), car dealers, and repair shops will become less necessary. Other jobs might pop up to replace them, but it is uncertain what they would be.

Widespread adoption of self-driving cars might just be another techno fantasy that won’t ever come into being. After all, vehicles powered by fuel cells (or hydrogen) or jet packs never came to pass. But autonomous cars are different because they can arrive incrementally. Cars already have cruise control, anti-lock brakes, lane sensors, and parallel parking assist. Soon private cars will be able to operate autonomously but only on highways and only when you choose to turn it on, then on city streets, and then eventually as Car2Google. Early adopters could be freight companies, then city buses, and then courier services. Unlike other technologies, it can function without everyone using it or a massive new fueling infrastructure in place. Self-driving technology can be layered onto existing cars using existing streets.

The public policy response to self-driving cars depends on how likely they will come into widespread use and how soon. Taking the pessimistic view, if this will never be more than a niche phenomenon, we should let the geeks play with their toys and ignore them. Driverless technology might find use in long-haul trucks or trains, but it won’t change cities in any significant way.

But what if we are only a decade away from being able to cheaply add self-driving navigation to any vehicle and on-demand automated car services suddenly take off? If that happens, today’s cars-first development policies will have been catastrophically wasteful. For example, if fleets of self-driving vehicles take hold, ubiquitous local laws that require each building to have its own oversupply of free parking will have wasted billions of dollars and vast acreages of urban land. Just so, in a future with fewer vehicles traveling through the system more efficiently, committing ourselves to multi-billion dollar roadway expansions is an enormous mistake. Buildings, parking garages, and highways are 100-year decisions. There is a built-in inefficiency with mass motoring, where every adult owns an expensive, space-consuming piece of capital equipment that sits idle 95 percent of the time. Our grandchildren may look back on this era as a short-term aberration, solved by technology.

The timing of self-driving cars and their possible impacts are uncertain. Nevertheless, the future leans toward transportation as a shared service. For decades we didn’t have the technology to allow one car to quickly and easily serve the needs of many people. Despite the inefficiencies of everyone owning his or her own vehicle, rules and cultural biases made the idea of car sharing seem not only impractical but even contrary to human nature. But we now have the technology; Zipcar and car2go have proven the market; investors see opportunity; and change is coming very quickly.

Naysayers have questioned the viability of car sharing since 1998, when there were only four car-sharing vehicles in our hometown of Portland. Today there are more than 1,100 here, with a few more added every week. Car sharing hasn’t stalled or died; in fact, it has exploded in popularity. Adding self-driving technology into the mix could make the next phase of car sharing truly transformative for both users and cities.


Ben Schonberger is a land use planner at Winterbrook Planning. Steve Gutmann is a consultant with a keen interest in car sharing, bike sharing and other forms of networked transportation. Both live and work in Portland, Oregon. Jade Chan edited this article.

We are a community-supported resource and we can’t do this work without you!



  1. Daniel Malarkey says:

    Great article.

    One of the big benefits of this technology is its potential to accelerate the adoption of electric vehicles. Users won’t have “range anxiety” because the electric car will matched itself to a trip request within the range of the current charge. Vehicles now generate over half of the greenhouse gas emissions in Washington so accelerating electrification is vital to meeting our climate goals.

    Another important win is the potential to reduce highway congestion and avoid costly additions to road and highway capacity. Self-driving cars can travel much closer together at speed and can thereby increase the effective capacity of roadways. If a meaningful share of peak period trips are made by self-driving cars we may signficantly reduce congestion and avoid expensive highway projects.

    If we get the public policies right to engable this technology to deploy at scale when it is ready, the region stands to reap big benefits: lower-cost travel with less congestion, lower emissions, and fewer costly highway projects.

  2. David Sucher says:

    An excellent article.

    One thing I wonder about is whether the possibility of the driverless car (D-car) should influence public choice of buses versus fix rail?

    I would think “Yes.” If indeed the driverless car is in the foreseeable future then expansion of Sound Transit (beyond the starter system we have planned now) as a rail-based system should be put on hold.

    Shared driverless cars will offer service which cannot ever be matched by “group transit” in trains. It makes much more sense to emphasize buses since the investment is lower etc etc

    • portrait Steve Gutmann says:


      I’m not sure whether/why our article’s driverless car scenario points to a future with less “group transit.” In fact, I would argue that it is equally likely to result in MORE demand for “group transit.” Here’s why:

      CarSharing trips are charged incrementally. Every trip has a transparent cost that is “fully loaded” (i.e. it includes the cost of the vehicle, parking, insurance, fuel, etc.) This is very different than trips via a personal car, where insurance, depreciation and other costs are fixed/sunk (and therefore opaque to the owner), so every trip has a low perceived cost (a few cents or dollars for gas). In other words, carsharing FEELS expensive, relative to driving an owned car, on a per-trip basis. This is why people who use carsharing services drive so much less than car owners.

      I believe that driverless cars (and driverless busses and trains) will, by making the actual cost of every kind of trip more visible, demonstrate that mass transit (again, once the cost of drivers is removed) is in fact more cost, energy and space efficient than taking a personal car.

      Driverless cars will also make it really easy to bridge the “last mile” to and/or from a public transit station.

      For these reasons, I think that under a driverless carsharing scenario, car ownership will fall and overall demand for public transit trips will actually increase.

      • David Sucher says:

        Your reasoning is interesting.
        And I guess it all depends on costs.
        We’ll just have to see.

        For example, car sharing services may NOT charge by the increment — i.e. you’ll get a monthly pass for so many miles. Who knows? Plus one would assume that there will be surcharges and discounts to reflect different times, days of the week etc etc. Your reasoning is interesting but it’s too soon to be sure.

        Driverless car sharing is so full of possibilities that it is exceedingly difficult to know too much.

        But personally, what I do believe is that most people
        • like to be able to move around, to see the world
        • with autonomy and privacy.

        That’s one of the reasons which I would personally invest in driverless car sharing long before I’d invest in subways.

        But who know. We’ll just have to see.

      • David Sucher says:

        But in any case your article was extremely good and thoughtful.

      • Wells says:

        I’m sorry, but the skeptic in all of us should find the electronic disruption impact objectionable in too many accident-prone ways. The Smart Car that ain’t electric, ain’t smart. This statement reintroduces the unfinished discussion of the question: Why BEVs are SmartCar-size suited vs PHEVs for compact, mid-sized family, multi-passenger ParaTransit vans, mid-size work rigs?
        60mpg SUVs. 100mpg compact coupes.

        Why do chicken coupes have 2 doors?
        If they had 4, they’d be a chicken sedan?
        There’s a metaphor in there somewhere.

        Self-driving cars does not address the immediate investment need: PHEVs & BEV mini-do-wop-a-be-bop urban hop around coocoocars.

  3. Matt the Engineer says:

    (has a terrifying daydream of autonomous bedroom cars designed for very-long-distance commuters out to new super-large homes built on cheap land in the middle of nowhere)

    The idea of autonomous car sharing isn’t really that big of a change from the current Car2Go model in Seattle. Every time I’ve needed a Car2Go I’ve found one within a few blocks. The benefits will really come to less dense neighborhoods in the suburbs that don’t have good car-sharing options. Will this be a plus or a minus for the envoronment? I doubt many suburbs will be built without garages, and it wouldn’t matter much if they were. And the new temptation to drive instead of take transit might be strong, as it seems like the strongest disincentive for driving into the city is the cost of parking.

    • portrait Ben Schonberger says:

      You could be right. But I don’t think the shared fleet model works in low-density suburbs. Like Car2Go, it needs a limited service area and a certain density of cars. However, people could add on full autonomy to their personal cars if it became cheap enough. Here is what I don’t know: people’s daily tolerance, in minutes, for sitting in a car. More if it’s driverless, and they can sleep or work, but it isn’t unlimited, I suspect.

    • Fnarf says:

      That’s the first thing that comes to my mind, too — three-hour commutes, working on other things while your car drives you in from the distant suburbs. Or, more likely, out — in the future we are approaching now, the good jobs will all be in exurban campuses while the richie-riches who hold them will live in ultra-expensive central cities.

      It’s interesting that proponents interpret “died in the legislature” as “legal restrictions falling away”. Liability is a huge issue, due to the way things work in our society. Even if self-driving cars reduce accidents by 99%, the family of the first person to get killed in or by one is going to sue the municipality for megabucks, and they’re going to win.

      I love seeing this “$9,000″ figure as well. Admittedly, I don’t drive very much, but I spend closer to $1,000 a year on my car, including gas and insurance and even repairs. I guess if you’re a lunatic and buy a new sports car every year and drive it 20,000 miles or something. Really, the biggest determinant of whether I drive or not is (a) is it to far or too difficult to bike or (b) price and availability of parking. Car share is dependent on the last of those as much as driving.

      In short, I’m a skeptic. Not just of self-drive cars, but of the very idea that the problems our cities face is technological or amenable to technological solutions. We have a housing crisis and a wealth crisis and a jobs crisis and an infrastructure crisis, not a tech crisis.

      • portrait Ben Schonberger says:

        On your liability point, my guess is that eventually legal and insurance interests will come around. Insurance companies look at pooled risk, not the isolated disaster. People are dying in car accidents right now, by the thousands, and lawsuits are being filed. “Better than a human driver” is a really low benchmark. If self-driving cars cut accidents and reduce insurance claims, the logical next step for insurers is to slash premiums on people who have self-drive cars and raise them on the laggards who don’t. It seems that the few big lawsuits that happen when the technology fails will be counter-balanced by all the lawsuits that never occur because of the reduction in crashes.

  4. Kevin Mclaughlin says:

    Great article. Given the ‘success’ of car2go, I’m still not sure about the the enviro / social benefits of urban driverless cars. But the opportunity is certainly there. My main point, however, is that one obvious component is missing from your story (and from all of the stories I’ve read that set up driveless cars to be the ‘magic bullet’ solution). What about the driverless minibus/jitney? While privacy is a key issue, throw in the Sidecar/Lyft phenomenon and we’re close to personal transit. (PS RichardGilbert.ca has written a lot about this in Canada.)

    • Matt the Engineer says:

      “What about the driverless minibus/jitney?” If your trip across town means going out of your way to pick up other people, you might as soon pay more for your own vehicle or just take the bus. That said, I’d think this would be palatable on long journeys, like the current vanpool market. I could see this taking people out of their cars – the cost savings of a vanpool but without worrying about leaving work on time.

      Ooh, or going the other way, what about the driverless tuk-tuk? Though I suppose the Car2Go approaches this in size.

      • Eric says:

        I think the driverless minibus could catch on. As has been pointed out earlier, driving seems more expensive when you’re using a car-sharing vehicle because all the costs are directly billed to you on a per-mile/per-minute basis . Full driverless mode is a natural extension of the car2go concept, but the high per-mile cost will still be there.

        What if the smartphone app you use to request one of these vehicles presented you with the following options:
        * Smart car, just for you, arriving at your doorstep in 5 minutes, estimated 10 minutes ride time. Cost: $5.
        * 15-passenger van, shared with other people taking similar trips, arriving at your doorstep in 5 minutes, estimated 15 minutes ride time. Cost: $2.
        * 15-passenger van, shared with other people taking similar trips, picking you up at the nearest arterial street corner in 5 minutes, estimated 14 minutes ride time (shorter because it doesn’t have to detour off the arterial to your house). Cost: $1.50.

        Yes, you might pay more for the personal vehicle taking a direct trip, but you might choose to pay less for a shared ride if the detours on the way aren’t very great.

        I could see something like this being very popular for commuting to work. Right now, carpooling is unpopular in large part because you have to coordinate your work schedule with the other people you’re carpooling with. The loss of flexibility isn’t worth the cost savings to most people.

        With a system like this you would be organizing carpools in a more spontaneous manner. How many Microsoft employees currently drive alone from north Seattle to Redmond every day? If a centralized dispatch app could match groups of them up for an ad hoc carpool based on when they’re leaving for work on that particular day, it could very easily eliminate a large amount of traffic from the floating bridges. The same vehicle could then load up with Amazon employees on the Eastside and take them over to SLU.

    • portrait Steve Gutmann says:

      I agree 100%, Kevin. A lot of transit agencies already use driverless trains, because they have dedicated rights-of-way. BRT is likely next. But once people (and insurance companies) get comfortable, I see no reason why shared transit wouldn’t use the same technology. And as Eric pointed out, the more people per vehicle, the lower the cost. Remove the cost of a paid driver (as we’ve done with personal cars), and the overall cost per passenger mile becomes a much more significant factor in what’s cheapest.

      IMO Eric’s logic re. pricing makes a lot of sense. If this technology takes hold, driverless cars will likely be the most expensive of various driverless transit options.

      • asdf says:

        “I see no reason why shared transit wouldn’t use the same technology.”

        Unfortunately, I do see one very important reason – bus drivers’ unions.

  5. Arthur says:

    Interstate trucking could be another industry impacted by driverless vehicles.

    “Australia’s biggest mining companies plan to add more driverless trucks to their operations … ”

    Australia’s big miners add more driverless trucks

    • portrait Ben Schonberger says:

      Absolutely. Those long, boring stretches between cities are ripe for automation. Perhaps there is an interim step where human drivers navigate out of the cities and then turn over the long hauls between cities to self-driving trucks.

      • David Sucher says:

        In fact my bet is that we will start with the Interstates and other limited access highways. Higher speed but so much simpler for the computers because so many fewer inputs.

    • portrait benschon says:

      Here’s a piece on the future of driverless trucking: http://www.strategy-business.com/article/00176?pg=all

  6. Ron Kinch says:

    Amazing. Economics and demographics screaming a demand to supplement other mass public transit with existing roads and reduce the number of vehicles, pollution and waste that we currently create plus the dangers of drivers who should not be driving putting others at risk.
    It’s a win win and would make moving people a more sustainable option.

  7. Jeremy says:

    “Cities could reap significant benefits: fewer parking lots, more relatively dense, walkable neighborhoods, cleaner air, and reduced carbon emissions.”

    Possible right now, and without the extravagant carbon burn rate demanded by the atomic, err, no, flying, err, no, self driving cars. But walking is hard work, and humans are lazy, and we’re fracking all the fields like the Red Queen runs, so…

    Predictions, assuming this self driving car thing actually pans out: loss of tax revenue, increased debt and reduction in savings as the job title “driver” quickly exits the payrolls of cash-optimizing companies, and all these folks retire or retrain for one of the many new jobs we’re creating to … err, no, automating away.

  8. Paul Godsmark says:

    Great blog. I have some observations:

    The Google self-driving car (SDC) has done over 500,000 miles of testing on public roads – 96,000 miles without safety critical intervention. On DriverlessCarHQ (now on Facebook) we estimate from photos posted on the internet that they have at least 32 vehicles and are probably averaging in excess of 1,000 miles testing every day.

    Bills passed in Nevada, Florida, California and District of Columbia. Bills in process in 15 other US states.

    ‘Less traffic’ – Earth Institute (EI), Columbia University study suggests that each autonomous car could replace 6 conventional cars.

    ‘Costs savings’ – EI study predicts average person could save 40% of transportation costs by using full-sized autonomous car. This could improve to 80% savings if using ultra-lightweight electric autonomous vehicle.

    ‘Sprawl’ – I expect that we will get sprawl, which may be counter to the goals of Transit Oriented Development (TOD). I also expect more intensification in urban centers because of this technology as well. So both more sprawl and more intensification.

    ‘Less safe streets’ – this is the very opposite of the goal of Google and I don’t expect this to be the case. This technology is capable of seeing 360 degrees and tracks all static and moving objects in the field of view and reacts in milliseconds. It is being programmed to be the most courteous and skilled driver – we should expect the streets to be safer, if not then this technology will not be certified safe for public use.

  9. Paul Godsmark says:

    My comments Part 2

    Widespread adoption will likely occur as this technology can make its owners money. I explain this new paradigm in much greater detail on my blog. This will be the first commercially available autonomous robotics that is going to change the very capitalistic construct – I cannot stress how important this could be if I am correct. Given that by simply owning this technology you can send it out to work and manke money, and the fact that 1 SDC can replace 6 regular vehicles then effective market penetration could be very rapid indeed.

    Transportation as a Service (TaaS) is very likely if you consider that the business models of taxi, car rental, carshare and rideshare companies all converge and start competing for the same business. I have written a blog post on the inevitable rise of autonomous vehicle fleets that represent this TaaS model evolving.

    I go further and put the case that cities may even compete to be the first to implement autonomous vehicle zones that will allow this technology to operate at full efficiency. This could see significant improvements in quality of life for urban dwellers, reduced emissions, greater safety and more space for community development and promote active transportation (pedestrians and cyclists) in our urban centers.

  10. Wells says:

    I’m sorry, but the skeptic in all of us should find the electronic disruption impact objectionable in too many accident-prone ways. The Smart Car that ain’t electric, ain’t smart. This statement reintroduces the unfinished discussion of the question: Why BEVs are SmartCar-size suited vs PHEVs for compact, mid-sized family, multi-passenger ParaTransit vans, mid-size work rigs?
    60mpg SUVs. 100mpg coupes. Electric Sharecars a few blocks.

    Self-driving cars does not address the immediate investment need in basic PHEVs & BEV mini-beeboppaloobop urban hop around kiddycars.

  11. Sam Churchill says:

    Here’s a proposal for autonomous cars on Hayden Island, on the Columbia River, opposite Vancouver Washington. http://i5bridgecam.wordpress.com/selfdrivingcars

    One advantage of battery-powered, rubber-wheeled vehicles is that expensive bridges are not required. No overhead line for power. No tracks. No parking.

    They could offer safer, faster, cheaper connection to public transit while providing improved bike access.

Leave a Comment

Please keep it civil and constructive. Our editors reserve the right to monitor inappropriate comments and personal attacks.


You may add a link with HTML: <a href="URL">text to display</a>