There was a good post and fascinating comments thread on Human Transit recently about driverless cars and what impact they may have on transport planning in the future. Jarrett Walker blogged that he can’t really see driverless cars being as revolutionary as some people think, for a few reasons:

  • Many of the benefits from driverless cars (such as increased capacity of the road system) only arise when there has been a complete changeover from current ‘driven’ cars and it’s difficult to see a pathway towards that eventual outcome.
  • Driverless cars will still (although potentially to a lesser degree) suffer from the ‘space extensive’ nature of individualised transport options so may not be that useful for very high demand routes.
  • If driverless technology becomes feasible then why wouldn’t there be a huge shift to driverless buses as well, which could dramatically lower the cost of public transport provision.

He concludes:

Sure, driverless taxis might replace many lower-ridership bus lines, but wouldn’t buses become driverless at the same time? In such a future, wouldn’t any fair pricing make these driverless buses much cheaper to use where volumes are high? Wouldn’t there be a future of shared vehicles of various sizes, many engaged in what we would recognize as public transit? As with all things PRT, I notice a frequent slipperiness in explanations of it; I’m not sure, at each moment, whether we’re talking about something that prevents you from having to ride with strangers (the core pitch of “Personal” rapid transit) as opposed to just a more efficient means of providing public transit, i.e. a service that welcomes the need to ride with strangers as the key to its efficient use of both money and space.

As I noted earlier, the comment thread is interesting because a few of his questions are answered in quite a lot of useful detail. For example, a progression path from the current system to a future transport system based around driverless cars:

1. A car maker introduces a driverless model that essentially works as a souped up cruise drive. A driver is still legally required, but the car will drive itself when you toggle it into cruise mode. This model will be expensive, but it will sell well to rich people who don’t like driving. Liability will naturally belong to the person who is in the drivers seat.
2. As these cars become more and more popular and proven to be safe, old/disabled people will lobby for regulations that the person being in the drivers seat don’t have to have a drivers license.
3. As these are getting safer and safer, regulations for someone being in the drivers seat will fade. More upper middle class people will buy them to driver their kids, pick themselves up from the airport, and so on and so forth.
4. At some point, taxi companies/uber start to buy these cars because they are cheaper then paying salaries.
5. As the number of automated cars grow, cities realize that they need smaller lanes and move more cars per lane. A few really big freeways will start seeing automation only lanes.
6. The prospect of skipping congestions means that they will sell better, allowing for more automation lanes to be built.
7. Meanwhile, competition slowly forces down taxi/uber prices, making car ownership less desirable for lower classes, reducing manual cars on the road.
8. Car makers only make automated cars because poorer people are buying less and less cars, and well off people all demand cars that at least CAN drive themselves.
9. And we are in the future utopia already.

I’m not completely up to date on the whole driverless cars thing. Some obvious issues that come to light are things like legalities when something goes wrong and how, if not impossible to work around, it’s certainly likely to slow down implementation. This is highlighted by another commenter:

…every time a driverless car hits a child who darts in to a street after a soccer ball or plows in to pedestrians in a crosswalk will set the movement back. When people-driven cars do this, we can usually find fault (“they didn’t see the kid”, “they were distracted by their phone”) but when a computer does it, there will be no easy answers and people will call for the cars to be off the road.

I guess one big advantage of driverless cars is that if they really do stop perfectly to ensure they don’t run over pedestrians, it pretty much turns every street into a defacto shared space because the vehicles will always automatically stop for you when you’re crossing the road.

Share this


  1. That last bit you write is interesting. If driverless cars are going to stop for you if you walk out in front of them, then guess what, I’m going to walk out in front of lots of them. Bugger waiting when I cross the street. And hence driverless cars won’t be very popular since they are going to stop a lot in city traffic. Of course the odd non-driverless car is going to be a disincentive, but hey you put one leg out, and if the car stops walk, if not pull your leg back in. This I take it would give old style drivers many heart attacks, hence driving as well as being in driverless cars is going to be a worse experience than the current situation.

    My take on it is that the first places where driverless cars are rolled out city-wide are going to be seen as failures and the technology won’t be widely adopted.

    The real advances in car technology are going to be when we stop the folly of having to accommodate 1.5 tonne machines carrying only one person wasting valuable space wherever they move and don’t move.

    1. I was thinking the same thing, no longer will I bother to use pedestrian crossing, nor will I bother to use the motorway overbridge – bring on automated cars as it will finally return the city to pedestrians. It would also be great for those of us cycling, no need to worry about traffic lights and because the automated cars will be constantly stopped for pedestrians it will be easy to nip around them all. Of course this will make accidents pretty frequent as actual drivers likely won’t have quick enough reflexes to stop all the time when the driverless car in front suddenly and frequently stops.

      1. I predict gridlock – exactly the thing they are trying to avoid :-). What we will have to watch out for is law changes to enshrine the rights of automated vehicles over everything else. A little way down the track for sure but look what happened with motorways and arterials.

      2. The laws of physics will still apply to automated cars. If you wish to treat motorways as suburban streets that’s your lookout, but don’t expect any sympathy when you end up dead because you assumed that a computer could do things that ordinary drivers can’t.

  2. I could see semi driverless systems being feasible, where people switch to automatic on the motorway. With divided traffic, no non-traffic road users, standardized lane markings and a sealed system environment as souped up cruise control with steering and stopping distance would be relatively easy. Cars would communicate by radio frequency with those around them using algorithms that approximate bird flocks or schools of fish. Off the motorway it just gets so much harder and failure prone, and as noted above the need to fail safe would probably stop the system from working outside the very controlled motorway environment.

    It’s similar to driverless trains. Driverless metros in sealed tunnels have been around for over half a century, those on grade separated surface environments for three decades. But the leap from driverless trains in a closed loop system to driverless buses on regular city streets is a huge one.

  3. So, essentially, nothing to see here; not so much driverless cars as computer assisted safety control. Good. That should eventually have an impact on crash rates especially as Nick says on motorways. But it does absolutely nothing to reduce the disbenefits of auto dominated place. We still need to reduce the quantity of vehicles, however they are driven, in many parts of our city if we want it to thrive and improve.

  4. I struggle to see how driverless cars will handle parking – surely there will be some manual guidance required there. There is no way a driverless car could select a car park that was available, unless there was some kind of car / park integration service built.

    I think it will be a long time before your driverless car could drop you off close to your work (something bus users experience already) and go and find a car park. The progression point of allowing driverless cars with no passengers at all is just too greater leap, without massive technical infrastructure investment.

    1. I believe it’s BMW who have been trialling a self-parking car that can be sent off to find a kerb-side park once the driver exits (as part of their driverless car system). Cars that self-park are already on the market, including in NZ, so it’s a small step up from that to a car’s control systems identifying an available park and then using it. The current impediment is that the car must have a driver so fully autonomous parking isn’t legal even if the car could navigate it way around neighbouring streets.

    2. Cam, there are cars on the market in NZ at the moment that will parallel park themselves once a park is found, some VWs and Fords have this feature

      given that the touted benefit of a driverless car (apart from driver laziness) is a close following distance, I can think of nothing more unnerving for a drive than being tailgated by a car with no-one behind the wheel, wait, wasn’t that a Steven King story?

  5. Agree with this article and the parent. It is especially interesting that driver-less cars advocates don’t seem to pick up that buses and minibuses possibly a better fit for driver-less vehicles than cars are:

    1. Buses go down a limited number of routes, so you only have to survey part of an area (this is a big part of what google does) to start using them

    2. Buses are used for many hours per day (and part of the reason they are not used 24×7 is cost of driver wages) rather than 1-2 hours/day for private cars so a (say) $100,000/vehicle cost can actually be a net saving.

    3. Bus companies can buy and maintain hundreds of vehicles at a time for further “bulk” savings

    4. When the cost of a driver’s wage doesn’t matter then smaller mini-buses with more frequent services, or more varied/flexible routes make sense.

    Driver-less buses mean that that you get to move towards some of the advantages of PRT while still keeping the advantages of moving people in bulk that PT has.

    1. Precisely. Based on what we currently know driverless technology will give public transport more of a comparative advantage than it has now. Ticketing would be interesting though.

  6. Here is an alternative view of the future, less excitingly new-tech I know, but more likely because hey it’s already happened: Paris has seen a huge mode share shift from cars and buses to trains and two wheeled vehicles. It is important to note that France and Eurozone in general has very high fuel taxes and therefore expensive fuel. Paris also is expensive park to park in because, you know, they didn’t flatten it for lots of carparking buildings. And by the way the French have been nutty about cars before kiwis ever even saw one so it isn’t just some cultural quirk but a rational response to changing circumstances.

    I would like to humbly add that this kind of shift to existing technologies has been predicted before on this very site:

    1. I would like to note here that in New Zealand motorbikes pay more for registration than cars. A small sub 60cc scooter pays 395$ per year! That is more than a 12 ton truck pays. I pay 521.21 for my bike and 280 for my car.
      Casual motorbike parking is rare and many traffic light’s sensors don’t sense bikes causing bikers to go with red or spend their life waiting.
      All of this makes it not worth it for the average commuter to shift to 2 wheeled vehicles.

        1. well there are good reasons from ACC’s point of view, as it is acting like a private insurer. However maybe from a WEB’s point of view its not such a good idea, and maybe a government agency should be directed to take this into account? I don’t know but if a transport economist did some investigation might find that they should have lower levies because of congestion benefits.
          Note if this same narrow thinking applied to sports players most would be priced out of the game (rugby especially), however the wider health and social benefits of sport are probably worth it. Of course I’m assuming fair collection methods which aren’t really possible, but still makes the point.
          On the other hand the noise of scooters can be a real menace in urban environments, although I’m sure the new ones are much less mosquito like!

          1. ACC is funded in three ways. Funds for work related claims are paid for through income tax, vehicle crash claims are funded through a levy on vehicle registration, and non workers non vehicle (i.e. sports, houshold injury etc) are funded from the general government pot. You’d have to convince the government that rather than funding motorcycle crash related claims out of motorcycle registration they should be funded out of the general tax fund instead. That would be a hard sell, particularly because the decongestion benefits of motorcycle use is almost negligible, because it is so small. Effectively you have a very tiny proportion of road users that have a massively disproportionate injury rate and claim costs. I can’t see a government wanting to subsidise that behaviour.

          1. “I have to pay more because car drivers don’t pay attention and so cause 60% of motorbike crashes?”

            Yep, and this grates me immensely. I pay over $400pa to register a 600cc motorbike, for the privilege of having inattentive idiots in cars trying to kill me on a daily basis.

            It’s all about externalities. Motorcycles, if crashed, inflict almost all damage on themselves and their riders. Cars are very good at damaging others (pedestrians, motorcyclists, cyclists, properties). We all know who wins in a car v motorbike collision, yet we have motorcyclists (the victims in 60% of crashes) paying higher rego costs. I’m no economist but this seems a very inefficient way to price risk, with very skewed incentives. Not to mention my 150kg bike is causing negligible wear and tear on roads, I’m one less car using less petrol, taking up less space, causing less pollution and less congestion etc etc etc…

      1. We need more 2 stroke pollution like a hole in the head. And when I work in my Welly city office I sit there with eaqrplugs in for all the noise of the sirens, car alarms, idling trucks (WhyTF must they idle?) and yes the shitty little sound of 2-stroke buzz-bozes. There are already too many 2 stroke scooters on NZ roads. You can taste them when you’re a pedestrian. Go electric and ban the 2 strokes I say.

        1. 2 strokes are being phased out in favor of 4 strokes. Modern european 50cc scooters (not the nasty overpriced chinese stuff you find in nz) are mostly 4 strokes and can do about 50km with one litre of petrol. 5 times as much than a car in rush hour. They take up no room in the road or for parking and are almost silent.

  7. Another scenario: if driverless cars automatically stop for pedestrians or any other moving ‘obstacles’, it will negatively affect traffic flow in cities. As there’s nothing more sacred than traffic flows, there will be calls for removing these ‘obstacles’ out of the equation. Think fenced-up roads and grade separation on a grand scale.

    NZ Herald, June 2048: “The Minister of Transport officially opens first invisible forcefield along Dominion Road, the first of 7 Forcefields of National Significance to be activated. «This forcefield will allow a steadier flow of traffic by reducing the number of automatic vehicle stops. In addition it will significantly improve both road and pedestrian safety.»

  8. In a country with an average vehicle age approaching 13 years, it’s going to take several decades before the entire NZ fleet is automated and, consequently, all the wondrous benefits forecast can be seen. NZ will be somewhere between stages 1 and 3 for at least 20 years, maybe with some overlap into stage 4. I doubt stage 9 will be reached within the lifetimes of any of the current Cabinet.

    We should certainly not be running current transport policy with any eye to driverless cars, for reasons not least of which is that it’s incredibly unlikely that they will be on NZ roads within even the next five years.

  9. If you have the time, the entire comments thread of that post is worth a look. The states needed to transition through are much more easily addressed than he assumes – well before you get fully robotic cars you move through glorified cruise control, and that creates a different set of needs from the system.

    But we are right to be skeptical of an easy or straightforward path.

  10. Driverless cars would surely put taxi drivers out of business? Cause everyone could drive to town and the car is your sober driver? But one good thing about driverless cars is Aucklanders would finally learn how to merge…..

  11. The main issue seems to be most of the benefits don’t accrue until everyone is driving a driverless car. In the meantime add-ons that are the start of this technology will be included with high end BMW’s etc, just as what we are seeing with automated parking.
    Also in bumper to bumper traffic they don’t really offer an advantage, as the capacity increases from driverless cars come from the shorter following distances.
    Interesting reading the economist article linked to by Jarrett. Lots of the benefits are supposedly in capital costs. But thats what rental cars, and car share schemes are for right? Have they taken over the world, well no.

    1. Interestingly there was a noticeable boom in new car rental businesses in inner Melbourne that seems to have gone hand-in-hand with the increase in density there over the last decade. I spoke to a couple who live in an apartment without a car who have an account with one these new agencies as they have a bach on the Gt Ocean Rd that they visit monthly. It works out cheaper and better for them to hire a car for these trips than to own one and have it sitting in a car park most of the month. The rest of the time they cycle, use Transit, and occasionally taxis.

      1. You might be interested in this link –
        I believe this company works in the same way as the people you mentioned in Melbourne.

        Friends in Sydney, going car-less in Glebe, hire rental cars for weekend excursions, as do a number of families living in this neighborhood. During the week they walk, bike, scooter, bus, tram or train…all of which is convenient to their home. There is just no space to conveniently park more than one car per household, unless you want to remove the very nice living space from the back of the terraced homes which are common to this suburb.

        And that is the heart of the problem with an over reliance on cars whether automated or not – they support a lifestyle that is detached, and suburban or rural in nature – and disrupt other more “communal” or urban” lifestyle options.

        1. Goget is the same principle as cityhop in Auckland modelled off successful car-share schemes like ZipCar in the US or mobility in Switzerland. However, I think what Patrick is talking about is people renting cars for 1-2 full days i.e. the weekend. Car shares are better suited for short hourly trips as the price usually ramps up once you start taking it for 24hrs or more and they usually limit the miles you can drive. If you want a car for the whole weekend it is probably cheaper to simply rent a car for 2 days from someone like Avis.

          A quite neat scheme running in the US allows you to register your own car and have people rent it by the day or hour (

          1. City hop is pretty good, $15 an hour, $30 from 6pm to 8am the following day, or $75 for 24 hours all inclusive (fuel, insurance, the lot).

            That’s still quite reasonable for longer hires.

  12. Certainly agree they are very useful for urban dwellers living in cities with auto dependent suburbs, and have been growing in popularity.
    Rather was attacking the thought of electric car proponents that capital cost savings will provide a boom for driverless cars, because those capital cost savings are available now through ride share schemes. However the ride share schemes still have a tiny market share.

  13. I can see benefit in driverless travel but only on say a motorway, once on the ordinary road a driver surely must take over there are too many things to detect I suspect the sensors would not pick up. Will they only detect metal? If so pedestrians and cyclists etc. will not come into the equation and be severely at risk. Many bicycles and motorcycles are made of carbon fibre and do not register for traffic lights will they be included in the car’s “vision”??

    There are too many unknowns to bring this idea to fruition, it’s frankly nutty!!

  14. There’s been a long-running (and probably irresolvable in the short term) debate amongst “innovative” transport enthusiasts about whether “personal rapid transit” or “dual-mode” systems (similar to what Nick R describes where a human driver controls the vehicle until it enters a controlled system like a motorway) are the way of the future. Compared to either of those solutions, full “driverless cars” would seem to have some signficant safety hurdles to overcome before they become a viable alternative.

    Addressing the issue quoted in the original post “Why would you want to a driverlass car compared to a driverless bus?”. Current public transit systems depend on grouping people moving from similar origins to similar destinations in order to be cost effective. Even a driverless transit system like Vancouver’s Skytrain is limited in its reach by the capital costs of construction. What PRT, dual mode and driverless cars have in common is that your desired most of transit will take you to straight to your specific destination without making compromises to attract a sufficient ridership. I’d love it if I could rock up to my local bus stop, find a vehicle waiting there to pick me up, which would deliver me to my stop in the city without making any other stops along the way.

    I find PRT attractive because the vehicles only have to be designed to cope with the known contingencies of a closed loop system, rather than every possible eventually a car might encounter. You can have one central control system for the whole network, rather than thousands of autonomous control systems negotiating with each other (some of those control systems being software, and others being human drivers). And it gets around the “What do you do with the vehicle when you’re not using it?” issue, by recycling the vehicle into the fleet for someone else to use (instead of sitting in a carpark 95% of the time).

    I just have this suspicion that driverless cars may end up like fuel-cell powered cars – technology that’s forever five years away from being viable on a commercial scale.

    1. I’m not sure what sort of PRT you are talking about here. The PRT used at Heathrow still cost 3-5 million pounds per km (
      What you are talking about doesn’t sound much like Public Transport at all, with most of the benefits disappearing too.
      The idea of having tens of thousands of these things descend on the CBD though is crazy.

      However best use of this technology could be for Paratransit in low density suburban areas, where driverless shuttles transport people to the nearest Rapid Transit Stop/town centre, whereby the could access the PT network to whiz them easily around the city. Then get benefits of closed network but having a limited route around local streets. Service would be online bookable, and most efficient route would be determined by the system as it receives the bookings. I still see any technology like this being decades away for large scale use, due to the many issues to be overcome by driverless cars.

  15. Could one implementation pathway be that certain roads are designated as legal for driverless operation, particularly motorways and certain rural roads ? The driver could then legally do other things, such as sleep, read or work on a tablet, and the risk of accidents with pedestrians and animals deemed so low that “they are at fault”. Then at the end of this road designation, control would be handed over to the driver and it becomes automatically assisted driving, where the driver now has legal liability.

    One of the current advantages of PT is that people can do more productive tasks than simply controlling a car. Semi-automated cars would reduce this advantage.

  16. I enjoy using active cruise control (accellerates and brakes to maintain distance/speed).
    & full autopilot is quite possible without external sensors – Google’s fleet of driverless cars have been driving around California for several years and now have “licenses” (though the LIDAR gear onboard costs $200k). Mining vehicles and ships also work on autopilot, due to the complexity & precision of manoeuvres required (also efficiency).

    NZ does have very inefficient driving habits, so even a simple active cruise & lane control could squeeze cars together, adding perhaps 30% capacity to our network; but it would need to include recording cameras/gps, so privacy may be a barrier for some.

    As for the sensors….Rome recently implemented a Smart Grid – power lines that are linked to a network that monitors consumption/demand/production – it really improves efficiency. Extended to it’s “phase 9” form, a city smart grid could encompass a wide network of sensors; power, lighting, crowd flow, traffic movement, road signs & lights, glass-break sensors (accidents & crime), alarms, weather & environmental monitoring, parking, cells & RFID, waste…and car parking. . It’s big brother stuff but thinking “Google Maps on steroids” and uploaded into vehicles (same as TomTom’s Traffic IQ does), then we are in business. Once a network (wifi or fibre) is established, adding sensors is really quite easy and not expensive. I’m told Council’s new I.T. system will have this capacity.

Leave a Reply