Cars & traffic

Jun 23, 2014

Are driverless cars only a generation away?

It increasingly seems it's not if we'll be transported in driverless cars but when. There're lots of potential benefits but realising them will be hard. It's likely there'll also be some downsides

Alan Davies — Editor of The Urbanist

Alan Davies

Editor of The Urbanist

Heading towards autonomous vehicles - where manufacturers are currently at with driver assisting technology (source: Bloomberg Businessweek)

Fairfax Media reported a claim last week that self-driving cars could be “mainstream in Australia within a generation” and could slash the road toll by 80 per cent (Driverless cars closer than you think).

Risk consultant Mike Erskine told a transport conference in Victoria the technology necessary for autonomous cars to operate on public roads is already well advanced. He said it will:

come into use in some form over the next 10 to 15 years and engineering controls, when developed well, usually have something like 10 times the reliability of what a person can do for a task. Experts know human error is to blame for about 95 per cent of collisions on the road

While advanced technologies are already appearing on the market (see exhibit), a generation sounds awfully optimistic to me. I don’t doubt, though, that driverless cars (a.k.a. autonomous cars) are on their way and will have a profound impact on the way we plan and manage our cities (1). Google expects to release its autonomous vehicle technology in five years and GM, Audi, Nissan and BMW all expect to have driverless cars on the road by 2020. Even if we add on an extra 20 years for hyperbole, it’s an idea whose time seems close – certainly within strategic planning time frames.

Yet even though they ostensibly look up to 40 years ahead, neither Sydney’s (still) draft metropolitan strategy nor Melbourne’s (finally) finalised strategic plan, Plan Melbourne, seriously discuss how autonomous cars might affect their development in the long-term (see Should strategic planning ignore the future?). From today’s perspective there are still big questions about the outlook and the ramifications of autonomous vehicles, but it’s a very big call to assume by omission that autonomous vehicles will have no impact on the shape and operations of cities by the middle of the century.

This is an important issue but it’s also a complex one. There’re lots of claims, both positive and negative, that warrant close scrutiny. I’ve discussed autonomous cars a number of times before, so I want to revisit what I said a bit over a year ago in Could driverless cars reshape our cities? (2)

Transition and implementation


A lot of the discussion on driverless cars focuses on the considerable problems of implementation, particularly the transition period during which human-controlled vehicles are likely to share road space with machine-controlled vehicles. This could take decades so there’re bound to be serious problems. Some of them will be technical but most will be political. What would the reaction be, for example, the first time an autonomous vehicle collides with a pedestrian?

I think there’s a parallel here with the introduction of cars at the end of the nineteenth century. Back then it wasn’t obvious cars would succeed on the scale they ultimately did. They were expensive to buy and operate for all other than the extremely rich. There was limited supporting infrastructure such as fuel stations and all-weather roads. The vehicles themselves were mechanically unreliable, difficult to control and operate, and unsafe for occupants. They were seen as a serious threat to pedestrians and horses as they were capable of what must’ve seemed incomprehensible speeds.

Moreover, there was active opposition to cars. As Aaron Wiener notes in the Washington City Paper:

….in the early days of the automobile, when the technology itself was being questioned and few rules existed to govern traffic and parking, there really was a war on cars. Driving could get you arrested in some places, whacked with stones in others, and actually shot by gun-wielding police in at least one. This wasn’t a philosophical debate over parking or bike lanes. It was a real, knock-down, drag-out battle.

I don’t think autonomous cars will provide the same quantum leap in mobility and productivity that cars and trucks offered back in the early twentieth century, but it seems likely they’ll nevertheless offer a compelling, even irresistible, proposition.

Potential benefits


Subject to the technical problems being overcome, the positive view of fully autonomous cars is they have the potential to lower the capital and operating costs of travel, increase speeds, and make time spent in-vehicle more productive and more comfortable.

Since intelligent vehicles are much less likely to have accidents than human controlled vehicles, they could be considerably lighter. They could consequently be cheaper to make, use less fuel and be more amenable to alternative power systems. They could be faster too because inter-connected computers can manage higher speeds better than humans. They could also manage congested traffic conditions intelligently.

Since they won’t require a driver, passengers could spend time working, reading, drinking, or sleeping. People who currently can’t drive – like the young, the elderly, the infirm and the drunk – could enjoy greatly enhanced mobility.

Autonomous cars could spend less time searching for parking spaces; they could drop passengers at their destination and then take themselves to the nearest car park. Probably the most cited potential benefit is greater road capacity. Because autonomous cars can travel faster and closer together, roads could take more vehicles, delaying the need for additional infrastructure.

Some of the biggest benefits could potentially come from car-sharing. Rather than own one or more vehicles that sit parked most of the time, households could summon a rented vehicle as and when needed, much as they currently use a taxi. And of course as Mr Erskine says, there could potentially be many fewer accidents due to eliminating human error.

Car-sharing isn’t an inevitable way autonomous cars would be deployed but they make the possibility plausible. Sharing would save households money and lower income travellers could avoid the capital cost hurdle of car ownership. For the society as a whole, sharing would reduce the number of vehicles that have to be manufactured and the number of parking spaces required in sought-after locations. Since the majority of trips involve only one or two persons, the average size of the vehicle fleet could also be smaller and lighter.

Autonomous cars won’t eliminate traffic congestion. Nor will they obviate the need for public transport in dense locations, but they could make it cheaper by eliminating the expense of drivers. That would also be true for trucks, delivery vans and buses (there are already a number of rail systems with driverless trains).

Potential downsides


But there could also be problems. If travel is cheaper, faster, and more comfortable, it lowers the cost of ‘driving’ ever further. People can consequently live further from work and other key destinations. That’s how trains and trams, and subsequently cars, facilitated suburban sprawl. Easier travel provides a private benefit – it increases the range of housing/locational choices available in terms of space, amenity and affordability, as well as easing pressure for redevelopment within established areas. But as we know from the history of urban sprawl over a century and a half, it can also impose social costs.

Autonomous cars could increase travel in other ways too. Even trivial decisions could lead to more travel – for example, some households might order the car to do multiple automated shopping trips on a just-in-time  basis rather than do all shopping in one trip. Those whose travel options are currently restricted because they can’t drive would likely travel more; that’s mostly a private benefit but could have a social cost.

Another issue is the potential for sharing autonomous cars might be more limited than much of the discussion assumes. Many travellers might prefer to have their own driverless car. They might not be prepared to wait for a shared vehicle to be dispatched or they might like the customisation potential a dedicated vehicle offers.

Even if the technology is only a generation away, fully achieving most of the positives rests on the assumption that all cars are autonomous and are supported by complementary infrastructure and laws (3). However it’s likely the potential benefits will be seriously compromised while ever legacy driver-controlled vehicles remain in the fleet. The transition to a fully autonomous fleet will likely be a challenging period politically, but policy-makers should be thinking now about the implications for the way cities might work in the future.


  1. I’m assuming Mr Erskine was talking about a population generation, not a technology generation.
  2. This article was previously published in a slightly different form at The Urbanist on 22 April 2013 as Could driverless cars reshape our cities?
  3. That’s a key reason why “closed” systems like rail lines have seen the first implementations of driverless technology.
(Visited 11 times, 1 visits today)


Leave a comment

Your email address will not be published. Required fields are marked *

12 thoughts on “Are driverless cars only a generation away?

  1. supermundane

    ‘They were seen as a serious threat to pedestrians and horses as they were capable of what must’ve seemed incomprehensible speeds.’

    Incomprehensible? Strange when locomotives of the time were already capable of attaining speeds of over 140 km/h. I’d say that the speeds attainable by the first horseless carriages were quite comprehensible.

    As to the article. I’ll contribute two more potential downsides.

    A representative from Google recently discussed the technology in detail. At present the technology requires detailed 3 dimensional mapping of the streets right down to the height of gutters, islands etc. The Google cars work well around Mountain View, California because Google has successfully mapped around 2000 kilometres of roads in and around the Googleplex out of the 6.5 million kilometres in the Unites States alone. I suspect when we see driverless cars rolled out, they’ll be confined to inner-city hubs for a long-time before the reach remote country roads where driving will predominate. They’re simply a long way from being intelligent enough and I suspect that people are being overly optiministic about this technology based on the industry hype.

    Another downside is that this technology signals a ratcheting up of government and corporate surveillance and all the attendant potential ramifications that this entails.

  2. Tom the first and best


    Automation does not solve the fuel issue.

    Traffic induced by automated cars, fuel availability permitting, would counteract any congestion benefit from shorter stopping distances in congested situations.

  3. James

    Chris & Dylan,

    I still am not convinced. I think the temptation to utilise this technology will be very strong in almost all locations. Is it not urban areas which have the most to gain from autonomous vehicles (reducing congestion, lowering accidents)? They are also the areas where we need to protect amenity and livability. I see segregation as the likely outcome and total entrenchment of personal private transport for all eternity.

    Chris, your suggestion of strictly underground use will still have to pass a cost/benefit test. We know how expensive tunneling is and what you’re suggesting are underground freeway networks…

  4. Dylan Nicholson

    Tom, obviously you would have an override system, and frankly the consequence of someone being trapped in a car (remembering that windows can be opened etc.) are not nearly as bad as being knocked off your bike at high speeds by a door flinging open. If we have sensors and computers reliable enough to navigate cars through traffic, then I think we can rely on them to prevent doors from being opened unsafely.

  5. Tom the first and best


    While dooring is a concern, a sensor based system for looking the doors unless there is plenty of space for them is likely to led to people being trapped in cars and a lot of upset people not able to open their doors.

  6. Chris Hartwell

    Austin, consider that the most likely design will be where all vehicles in a given range are communicating their status and intent to one-another – large-scale peer-to-peer network. This would include emergency vehicles and would largely resolve the issue you raise about lines of traffic/temporary markings. I’d expect the vehicles to automatically decelerate if a potential road hazard, outside of the vehicle network’s control (child, animal, cyclist) were detected in the immediate vicinity.

    I’d also contend that with such a networked system, red lights would no longer be necessary – the vehicles would simply negotiate intersections in such a manner as to avoid lengthy delays and allow for near-continuous flow.

  7. Dylan Nicholson

    Austin, I’d expect that initially a licensed driver would still be required to be at the wheel, sober and awake. As long as the computer was capable of detecting that a situation was approaching where human invention might be necessary, I don’t see any of those as major issues. But eventually computers will even be better at handling those situations than any human could be. At least you didn’t bring up the example of having to decide between hitting one person on one side, or hitting two on the other. I’m curious if any human has ever had to make such a call.

  8. Austin M

    I still think there are a number of hurdles for the autonomous vehicle to overcome and am not aware on any autonomous vehicle being tested in what I would call really taxing environments.
    I often think about how an autonomous vehicle would cope in a major road works area. How would they go with blacked out or grounded out lines v.s. temporary and contradicting painted lines? With dirty roads and road signs? With poor delineation or direction? With counter intuitive operation? With random events like a traffic controller or a concrete truck blocking the road and backing into a driveway? An accident situation? An emergency vehicle behind them v.s. red-light situation? To cross onto oncoming traffic or mount a kerb and avoid hitting a child but not to do so for a cat?
    Making a vehicle understand how to operate to a set of rules is one thing but enabling it to understand how and when to break the rules is a level of capability that im yet to see a concerted effort or capability in. That needs to be overcome if we are to ever have truly autonomous vehicles as opposed to semi autonomous highway cruisers and parking machines.

  9. Dylan Nicholson

    James, I’d think the opposite – autonomous vehicles will be far more proficient at avoiding pedestrians than human vehicles, and as we gradually become aware of this, pedestrians (and cyclists) will learn to feel comfortable with such vehicles moving around them. I’d expect, for instance, they would be programmed to simply not enter areas with significant pedestrian traffic if some threshold for causing inconvenience or danger was reached. One technology that I would consider mandatory is that doors cannot be opened until the computer has judged that doing so won’t cause harm or alarm to anyone outside the vehicle.

  10. Chris Hartwell

    James, you’ll likely see far lower speeds in pedestrian areas. The key savings with autonomous vehicles in pedestrian areas will be more sanity in merging/lane changing, intersection negotiation while high-speed would be reserved for major thoroughfares.

    A second, although far more costly possibility, would be the construction of a second “ground level” that is exclusively pedestrian and provides access points to the lower ground level where autonomous vehicles operate.

  11. James

    I wonder if more segregation will be required between pedestrians and roads where autonomous vehicles are active and whether this will destroy gains made in recent times to claim back space for other modes? What will this do to our cities and spaces? I can’t imagine our societies allowing vehicles travelling at very high speeds without barring the proximity of pedestrians and cyclists – even if these vehicles are controlled by computers. It’s still physics to stop a heavy moving object.

  12. Tom the first and best

    Taxis and trucks will be some of the first vehicles, regulation permitting, to be automated because the costs can be offset by eliminating the need for a driver and the space a driver takes up. This would also allow cheaper taxis and this would reduce the need for car ownership.

    Buses are a little harder because of the revenue and passenger spotting roles of drivers. However these are not insurmountable (non-driver based ticketing and buttons at stops to tell the buses to stop) and it will likely be followed by driverless trams.

    They will also start filtering through the private vehicle market from the upper en, as professionals, whose time is valuable take them up.

Share this article with a friend

Just fill out the fields below and we'll send your friend a link to this article along with a message from you.

Your details

Your friend's details