driverless cars and the limitations of the “complete imagined future”

Note:  This old post is still useful whenever you see a "driverless cars will change everything" story, (this one, for example) and especially a "driverless cars will be the end of transit" story.  Abstract: The two fallacies to watch for in these stories are (a) the "complete imagined future" mode, which denies the problems associated with evolving the future condition instead of just jumping to it, and (b) the assumption, universal in techno-marketing but always untrue in the real world, that when the whizbang new thing appears, everything else will still be the same; i.e. that none if the whizbang thing's imagined competitors will also have transformed themselves.  This latter assumption can also be called the "everyone but me is a dinosaur" trope.

Richard Gilbert, co-author of a book that I've praised called Transport Revolutions, has a Globe and Mail series arguing for how driverless cars will change everything.  I will give this series a more thorough read, but just want to call out one key rhetorical move that needs to be noticed in all these discussions.  It's in the beginning of Part 4, "Why driverless cars will trump transit rivals."

With widespread use of driverless cars – mostly as autonomous taxicabs (ATs) – there could be more vehicles on the road because ATs will substitute for most, and perhaps eventually all, private automobile use as well as much use of buses and other conventional transit. 

This, and much of the discussion around driverless cars, is in the complete imagined future mode.  Gilbert describes a world in which the driverless cars are already the dominant mode, and where our cities, infrastructure, and cultural expectations have already been reorganized around their potential and needs.  

Some complete imagined futures are not necessarily achievable, because the future must be evolved.  In fact, the evolution of organisms is a fairly apt metaphor for how cities and infrastructure change.  As in evolution, each incremental state in the transformation to the new reality must itself be a viable system. We can think of lots of wonderful futures that would be internally consistent but for which there is no credible path from here to there.  

Driverless cars remind me a bit of the "wheeled animal" question in evolution.  No animals have evolved with wheels, despite the splendid advantages that wheels might confer on open ground.  That's because there's no credible intermediate state where some part of an animal has mutated something vaguely wheel-like that incrementally improves its locomotion to the point of conferring an advantage.  Wheels (and axles) have to exist completely before they are useful at all, which is why wheeled animals, if they existed, would be an argument for "intelligent design."

I will begin to take driverless cars seriously when I see credible narratives about all the intermediate states of their evolution, and how each will be an improvement that is both technically and culturally embraced.  How will driverless and conventional cars mix in roads where the needs of conventional cars still dominate the politics of road design?  How will they come to triumph in this situation?  How does the driverless taxi business model work before the taxis are abundant?  Some of the questions seem menial but really are profound: When a driverless car is at fault in the accident, to what human being does that fault attach?  The programmer?  What degree of perfection is needed for software that will be trusted to protect not just the passengers, but everyone on the street who is involuntarily in the presence of such a machine? 

Here's a practical example:  In Part 3, Gilbert tells us that with narrower driverless cars, "three vehicles will fit across two lanes."  Presumably lanes will someday be restriped to match this reality, but when you do that, how do existing-width cars adapt?  If you could fit two driverless cars into one existing lane, you could imagine driverless cars fitting into existing lanes side by side, so that the street could gradually evolve from, say, two wide lanes to four narrow ones.  But converting two lanes to three narrow ones is much trickier.  I'd like to see how each stage in the evolution is supposed to work, both technically and culturally.

That's one reason that I seem unable to join the driverless car bandwagon just yet.  The other is that claims for driverless taxis replacing transit amount to imaging a completed new technology out-competing an existing unimproved technology — as though that would actually happen.  

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. 

78 Responses to driverless cars and the limitations of the “complete imagined future”

  1. zefwagner January 3, 2013 at 10:15 am #

    Of course, we already have examples of the effect that driverless transit can have. Vancouver’s Skytrain can achieve amazing frequencies because the agency doesn’t have to worry about the high labor costs of drivers. The big danger will be that even when all trains and buses are able to be driverless, they won’t be due to powerful unions protecting drivers’ jobs. Considering Oregon still doesn’t allow people to pump their own gas, the prospects are grim for driverless transit. Still, we should try to encourage it.
    I’m sure driverless cars will make transit even less attractive to people in the suburbs, but when connecting dense urban centers there is little substitute for high-capacity transit vehicles. I think this transition will force medium-density cities to really decide whether to go for higher density or not. If they choose to focus on the driverless-car model, they will still need space for all those cars, so they effectively choose to stay at a lower density.

  2. Kyle January 3, 2013 at 10:44 am #

    Though driverless cars may sound nice in theory, we must take a step back and analyse the impact on our cities: If a person can travel the same distance in half the time, he will choose to live twice as far, intensifying the current sprawl. This is assuming of course, that roads will continue to be free. What worries me is what will happen in Wall-E. Cars will be severely different that today: No Ownership (just Car2go type rentals), cars will be small, and the difference with it from transit, as some cars will require sharing (even if they are only 2 seats but my guess is most will be one), driverless cars will be door to door.
    But this is flawed as our cities should not be dominated with things, no matter how safe, how clean, people do not enjoy living with things going 50-100 and even maybe 200kph in our cities. The fastest human can run only at 45 kph, and humans cannot adapt to share our cities with things going twice that speed.
    So I envision a city where bikes and walking replace cars, and driverless cars replace public (mass transit), but where people, not cars will be king.

  3. Ben January 3, 2013 at 10:59 am #

    Why is this issue so clear to you and me (and many of your readers) but not a lot of other people? All the techno wankers think that their dream future can just pop into existence with no baggage from the past. How can they credibly imagine these futures without some sort of intermediate transition plan?
    Not only do you have to worry about infrastructure issues (e.g. the 3-2 lanes) but also the personal economic cost. With many people still driving cars that are 15+ years old, and the economy in the tank for the foreseeable future, where exactly is the money going to come from for us to transition to driverless cars in a reasonable timeframe?
    As an example, we’re having enough trouble trying to get people to convert to hybrid and electric cars. And that’s been going (off and on) for 20 years now. Without government subsidies and the manufacturers selling them at a loss the adoption of hybrids / electrics would probably be much worse than it already is.
    On the other hand, we have known for centuries how to build efficient train transit systems. Why can’t we just improve our trains (and I don’t even mean HSR — just basic connectivity and frequency) to start off with… that’s something we know how to do and we can’t even do it (here in U.S.)
    Yet people think we’re going to convert to driverless cars? Riiiiight.

  4. Scott January 3, 2013 at 11:23 am #

    The idea of driverless cars scares me, and I am not particularly happy about the idea of betting my life on the ability of some Google programmer, whatever safety claims are being made. Such is the nature of humans: we trust our guts, not statistics or the people who hire the statisticians…
    That being said, all the “This will never work so give up now!” rants above – including the comparison of driverless cars to wheeled animals – strike me as ridiculous. The alternative view is well presented in The Economist, and essentially comes down to (in my synopsis): Sure, if you see Massive Total Change Overnight as the standard then it is not going to happen; but do you really know everything that is going to happen in the next hundred years or so, which is the real time frame to consider?
    http://www.economist.com/blogs/freeexchange/2013/01/innovation

  5. Eric O January 3, 2013 at 11:31 am #

    That conceptual leap to imagine in-between stages seems quite large. Perhaps we are over-wired to think in terms of binaries? It is hard to imagine that a steady state between extreme ends is not only useful but might actually be preferable.
    It used to be that kind of binary thinking was always challenged among urbanists (thinkers like Kevin Lynch were always sensitive to beneficial steady states and stepped adjustments). But today, ask an urbanist to try to imagine a bus transit experience that could be superior to rail. A conceptual leap that includes a bus in the picture is just not pristine enough. Thankfully, I suspect that the incrementalist, DIY urbanists that are thinking more about transitional stages than heavens now might actually discover preferable steady states by accident.

  6. Wai Yip Tung January 3, 2013 at 11:33 am #

    The no conceivable intermediate state of evolution argument is interesting. Richard Gilbert may very well be wrong. All of his specific predictions about the future of transportation may well be completely off.
    But evolve it will. While we cannot say for sure where will it takes us, I believe there is a good chance that automated vehicle will play a role, even a significant role in the future. I hope to see traffic accidents greatly reduced, since human are such poor drivers who are overly competitive, have limited perception and become wholly unreliable when under influence of alcohol. While three car traveling in two lanes may not be a feasible idea, there are other broad trend that is likely to emerge in the future, such as robots will take over our jobs:
    http://www.wired.com/gadgetlab/2012/12/ff-robots-will-take-our-jobs/all/
    So while automated taxi fleet may not arrive in the near future, I see plenty other promising applications. The idea of taking over long haul truck driving on freeway seems conceivable. A fleet of automated shuttles in a college campus can be really useful. I think this is a huge enabling technology and our future will be quite different from today.

  7. Lee January 3, 2013 at 11:39 am #

    I would think that the evolution is fairly easy to imagine
    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.

  8. Mike January 3, 2013 at 12:43 pm #

    I don’t have any doubt that autonomous cars are probably in our future. I suspect it won’t be any threat to transit for a few reasons…
    – Like car sharing, cheaper cabs will lead to more people opting into the car free lifestyle. They’ll take cabs when the need to and buses and trains when it’s cheaper.
    – Autonomous buses/trams would allow agencies to increase service and compete with autonomous cabs in dense cities where they’re more efficient. Fare collection would effectively be off board.
    – Autonomous taxis would allow transit to let go of unproductive peripheral service, route deviations, and some paratransit; and focus on increasing ridership.
    – Safety improvements with autonomous cars will improve people’s confidence walking in cities. Roads will be plowed better, etc.

  9. Ari January 3, 2013 at 12:46 pm #

    Very well-said. I’ve tried making these arguments myself and always get the “but it’s so cool! and Google made it!” argument.
    Another hurdle: a lot of people like driving cars, and won’t want to give up the feel of depressing the gas pedal and hearing the engine whir. Saying that everyone is going to jump in to driverless cars is making the assumption that people act rationally when they make transportation choices.
    Finally: 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.
    If we see widespread driverless cars in my lifetime, I’ll be surprised. Considering cars have made negligible technological progress in the last 50 years (no increase in overall speed or throughput per lane), I don’t see this changing wholesale any time soon.

  10. Neil21 January 3, 2013 at 1:21 pm #

    I was going to write a full response over on http://technooptimist.tumblr.com/ but there’s so much of interest here, I’ll just append.
    I haven’t seen Brad Templeton or Neil McGuigan mentioned yet. Here’s Brad’s roadmap http://www.templetons.com/brad/robocars/roadmap.html and Neil on insurance (among others): http://vancouverdata.blogspot.ca/2012/08/googles-self-driving-cars-are-going-to.html
    @zefwagner is spot on. The biggest risk is unions (political, not tech or capital) and transit is still required on arterials by the geometry of congestion.
    @Kyle sounds confused. He is labouring under the common misunderstanding that sprawl is the result of free market choices not policy. He imagines robotaxis would travel at high speed, which I don’t (same price and time as a transit trip, but private and door-to-door, is enough of a value proposition).
    @Ben completely misses the robotaxi market growing out of carsharing. To be fair, even guru Templeton seems to assume private ownership far too often.
    @Scott is scared by driverless cars, but presumably not by human-driven cars, skytrains or autopilot. Not a credible forecaster.
    @Lee provides a plausible path, akin to Templeton’s.
    @Mike is spot on
    @Ari re “people like driving” is bunkum. People also liked riding horses. Economics and infrastructure (safety/pleasantness/ease) determine dominant mode choices, not some independently arrived-at personal preference. Re accidents, robotaxis never make the same mistake twice, unlike humans, and they perfectly record the scene of the crime. Google’s biggest issue is making the cars the less timid: they’re programmed to slow if uncertain, unlike human drivers who often accelerate in panic.
    As an urbanist and environmentalist, I can’t wait for robotaxis. I see car2go and modo offering bookings that arrive at my door, and then cars that drive the whole way. I see EV startups Fisker, Coda, Tesla, Better Place, hiring LiDAR specialists and former members of Thrun’s team, to get fleets of their cars operating as robotaxis. I see mining companies, who already run autonomous trucks, partnering with UBC to trial made-in-BC robotaxis for mining communities with DUI issues. Just as Nevada changed their laws to attract Google R&D so should smart suffering BC towns.
    Automation is the killer app for electric vehicles, by the way, by enabling easy taxi/sharing (and so lowering the cost of access) and because they can go charge themselves.
    Electric robotaxis on demand undermine individual car ownership. Without the sunk cost of the car in the driveway, it’s much easier to walk for more trips. As behavior changes, so follows identity, and demand for pro-pedestrian-propulsion planning and streets will rise.
    Of course we don’t all switch to robotaxis overnight, and every commuter can’t take a robotaxi. Car2Go and Modo didn’t start in sprawling Surrey, and neither will Robotaxi Inc. But robotaxis work in lower density environments than carsharing, so it helps expand the market.
    The recent Uber debacle in Vancouver bodes poorly for our being near the forefront of the robotaxi world. Ditto our strong union culture bodes poorly for robobuses. This is a shame for taxpayers and travelers.
    I recommend Venkat on the Future Nauseous, and how we don’t notice the future when it arrives, because it comes via metaphor extension: http://www.ribbonfarm.com/2012/05/09/welcome-to-the-future-nauseous/ Ditto Whistlecars evolving out of Carsharing.

  11. al January 3, 2013 at 1:23 pm #

    Driverless cars will operate autonomously on highways first. There are fewer variables on highways than on streets. Taxis and car services might be a nice market, as no driver means possible round the clock availability (with late night fueling and maintenance). It might drive down taxi and car service costs. Another possibility is to combine the technology with multiple steered axles to create multiple unit buses that resemble trams or trains. Alternatively, large numbers of self driving vans (5-12 seats) could become another mass transport solution.

  12. NCarlson January 3, 2013 at 2:40 pm #

    On that improved cruise control note, I really have to agree. We’re still quite some distance from real automated cars, but a vehicle that can maintain speed and distance while staying in the current lane seems quite possible in a model year or three – I’d say we seem to already by at the point where the biggest challenge to that is integrating the camera’s in an acceptable fashion for a production vehicle.
    As an aside, streetcars and light rail seem like a potentially very important stepping stone. It requires all the sensing and response capabilities of automated rubber tired systems, but without a lot of the complexity that comes of having to steer the vehicle. My impression is we’re still a ways from acceptable operation in pedestrian heavy environments, but that it’s more or less the next step. Once that happens driverless streetcars are something of a game changer for the whole modern streetcar concept, and light rail vs light metro debates end up being entirely about local environment and service quality.

  13. Richard Campbell January 3, 2013 at 2:48 pm #

    It really isn’t a question of if but more where and when. Our current transportation system based on ownership of vehicles and driving, is shockingly inefficient in pretty much every way; energy, resources, land, time and economic. We are so used to the current system, we really forget how wasteful it is. Then there are the safety problems which have a horrible impact on the lives of many people.
    Shared driverless vehicles for moving people and goods have the potential to be so much more efficient in every way that there will be little choice not to adopt them.
    I expect they will be first adopted in one of the hundreds of new cities springing up in Asia or Africa. When designing a city from scratch, it will be much less expensive to design it around automated vehicles. Due to the economic advantages, existing cities will follow sooner or later. Like with all major transformations, developed cities are at a disadvantage due to their existing infrastructure and powerful vested interests.
    As far as the debate between current forms of transit and automated vehicles, who knows how that will turn out. It is a bit of a distraction anyway. Both are transportation as a service and they can be seamlessly integrated using technology. They can be complimentary in many ways.
    Nodal transportation is more efficient and popular the further the nodes are apart and the faster the service between the nodes is. Rail based rapid transit, high speed rail are more likely to be competitive. Local serving buses and streetcars less so.

  14. bradtem.myopenid.com January 3, 2013 at 4:29 pm #

    I will admit we are often guilty of talking about the consequences of a world with high robocar and robotaxi penetration. That doesn’t mean we don’t think a lot about the incremental paths that can get there, but it is harder to judge their effect on urban spaces because it is both hard to predict which path will be taken, and the steps will be short-lived — too short lived for cities and planners to have time to notice, study and act upon them before the world is different.
    One fairly clear gradual effect will be a reduction in the need for parking. Robocars are going to need a lot less parking, both because they can valet park more densely, they can park in “standing” locations or remote locations, but more simply because if they are robotaxis they will not park at all — they’ll go off and transport somebody else.
    Even a small robotaxi fleet starts having a small but visible effect in the parking lot. There are fewer cars in it. And this grows until it’s quite noticeable that the parking demand is less and we start finding that the lot that was sized for peak demand is never full any more. You start seeing this with only modest penetration. Then lot owners start wondering if they should repurpose some of that wasted land and start going to the city to talk about it.
    I have called this robocar oriented development, a parallel to TOD, and describe it at http://robocars.com/rod.html
    The effects on congestion are not as quick. I believe an interesting approach is moving people into smaller vehicles (which they would never buy but would happily ride in on short urban trips) and eventually half-width vehicles. Any serious penetration of half-width, 2/3rd length vehicles on the urban street begins having a noticeable effect. The half-width vehicles are still several years away, but well within the transit planning horizon which is often 2 decades.
    3 cars in 2 lanes is an interesting question. Turns out one of the problems of half-width (which is easy to pack into today’s lanes) is stability. You need vehicles that can bank, or only go at low speeds. 2/3rd width is more doable but as you note, doesn’t mesh with our lane patterns. To make it happen you need to find 2 such vehicles in 2 lanes, and have them let a 3rd vehicle enter the middle as they move to the sides. This has the issue of creating a barrier to traffic for those who want to pass. In light traffic that’s manageable but in heavy traffic it’s an issue.
    So I’m not quite sure if the 2/3rds approach works but I don’t see it as inherently impossible. The half-width approach has a clearer path — and a greater traffic benefit.
    What excites me about small vehicles is their huge energy efficiency. Half-width electric cars are super-green. It is possible to make them so that they use less energy per passenger-mile than any transit system (even East Japan Rail) and much less energy than U.S. transit systems. (Not that this is hard. The average bus in the USA gets about 4mpg and has 9 passengers on it, less than 4 0 passenger miles per gallon, which is worse than a solo Prius. That’s shockingly bad. And don’t get me started on the light rails, which are worse. Only the subways and commuter rails get a decent green score compared to cars and they can’t beat the efficient electrics over their whole cycle — though of course they beat them during rush hour, but the transit system is not just rush hour.)
    Small, light vehicles offer the potential for a much greener transportation system. But up to now, we have not found a way to get people to ride in small efficient vehicles even on solo trips where they are just fine. I think robotic taxis can start changing that. And yes, incrementally, not just all at once.
    And yes, people are designing small, light vehicles in the sub 100 wh/mile range which are crash safe even with crazy human drivers on the road. We don’t have to await a completely imagined future.
    In the 20-teens, you won’t see urban planners wake up, though. What we’ll see this decade is more private, such as the 2104 Mercedes and Volvo models (coming out in 2013) that cruise the highway steering themselves. Their main social consequence will be the fact that, like their ACC predecessor, they smooth out human driving patterns and don’t rubberneck, which helps with freeway congestion. You’ll also see auto-parking cars mid-decade that valet park in lots, but not remote lots. When the taxis come — perhaps in the teens, perhaps early in the 20s — we start seeing the bigger changes.
    But make no mistake about it. The “completely imagined future” is coming, and sooner than you think. Nobody can predict everything about it with perfection. But for transportation planners to ignore the key elements of it would be a serious error.
    I have made a summary of robocar potential changes which affect urban planning at http://www.templeto ns.com/brad/robocars/urban-plan.html . Yes, some of it is about the completely imagined future but I think all the issues need consideration soon.

  15. Ben January 3, 2013 at 4:39 pm #

    I’m sorry, but this wankery autopilot techno shit is pure fantasy. It can work for things like airplanes, sure, because they’re flying miles away from any obstacles. But cars? Are you people serious?
    That Google robot car may be able to drive safely on pristine controlled highway conditions, but it would be extremely difficult to make it work in most driving situations. There are simply too many variables to consider and too many subtleties of driving that computers simply can’t be programmed to deal with. Like jerk drivers on the road who cut you off, traffic that is coming to a sudden stop where it shouldn’t, objects flying across the roadway, and so forth and so on.
    This reminds me of a show that I used to watch on the Discovery channel back in the ’90s called “Beyond 2000”. It always showed some kind of techno magic that was just around the corner. One that I always remember is the car that was outfitted with computer chips and sensors in the tires so it could see obstacles on the road surface (rocks, potholes) and calculate the suspension reaction in real time so you’d get a perfectly smooth ride.
    It apparently worked great on a test track with the rocks and potholes, but they noted briefly that they still had to figure out how to get the system to tell the difference between a crumpled paper bag and a rock, or a smooth surface and a pothole full of water. Guess what? They never figured it out — because these things are difficult for computers to do.
    No amount of techno magic is going to save us from the poor decisions that our society has made over the past 50+ years. Buckle you’re seatbelts, because we are in for one hell of a ride over the next 15 years as cheap energy runs out!

  16. Ben January 3, 2013 at 4:47 pm #

    I meant “your seatbelts”, obviously — thanks autocorrect!

  17. Miles Bader January 3, 2013 at 7:29 pm #

    As, or more, important than “how” is “what’s this imagined future actually going to be like when/if we get there?”
    People in the 1950s imagined mass car-ownership and suburbanization to be some sort of utopian state, and consequently devoted massive resources to achieving it over ensuing decades—only to be confronted with the dismal and depressing reality of what those things are actually like.
    “Be careful what you wish for, you might get it” seems apropos…
    The sort of people I see championing driverless autos strike me as being classic proponents of the “tech fix”: that we shouldn’t worry about “problem x” because “technology y is on the horizon, and will solve everything [waves hands a lot]!” Needless to say, things usually turn out to be not so simple….
    This sort of utopian day-dreaming is in many ways a positive thing, but it becomes a very real problem when it’s used an an excuse to avoid the hard work of addressing immediate problems in favor of just waiting to be saved by an imagined future.

  18. anonymouse January 3, 2013 at 7:47 pm #

    There’s an important difference between a “self-driving car” and a “driverless car”. The former still has someone in the driver’s seat, they’re just not doing most of the tasks we think of as “driving” much of the time. The latter actually has no “driver’s seat per se”, and while the tech is mostly the same, there are important differences in other areas, and the former is likely to be common much earlier than the latter. The same thing happened with rapid transit: self-driving trains like the Victoria Line and BART and such started to be common in the 1960s, but fully driverless systems like Skytrain were not built until the 1980s.
    I suspect the first “self-driving” technology will be marketed as just an enhanced form of cruise control for use on highways, and will still require some human supervision. But it’s still enough of a safety and convenience improvement that it can make sense on its own, if it’s cheap enough. And given the likely improvements in safety, insurance companies will start offering discounts for it, which should help increase adoption too.

  19. MikeB January 3, 2013 at 8:43 pm #

    The biggest hurdle for driverless cars is that they can’t be jerks.
    If a self driving car is safe enough so that it will satisfy the liability issues for the manufacturer, it will be so gentile as to be ineffective. First, it will have to observe all traffic laws, no speeding, no running stale yellows, no ignoring crosswalks, no double-parking. Just that much will make it slower and therefore less attractive. The driverless cars will also have to be far more cautious, especially in urban settings. I can foresee traffic jams started because the self-driving cars will not nose into a stream of pedestrians crossing against the light, or because they insist on changing lanes completely to go around a pedestrian or cyclist moving slowly on the sidewalk.
    Self-driving cars may eventually be great for safety and for urban places, but I don’t expect they will ever become common because the benefits for daily driving just are not there.

  20. mike0123 January 3, 2013 at 10:44 pm #

    In mixed traffic, self-driven human-supervised cars are no more useful for transportation than human-driven cars. As long as they share space with human-driven cars, they must follow the same rules and they cannot achieve tighter spacing or higher speeds. As with human-driven cars, they can be used in car-sharing systems, they cannot be used while drunk or blind, and they make you look for a parking spot from which you have to walk to your destination. The cost structure changes little for the user.
    In mixed traffic, driverless cars are more useful than human-driven cars in some ways. They also must follow the same rules, though, so they aren’t faster or more capacious. They make door-to-door trips possible. They give mobility to the blind and drunk. They move parking to less valuable land.
    If driverless cars are shared, they reduce the need for parking and associated costs. If that cost is significant, car sharing may become less expensive than individual ownership for commuters and the pay-per-use model would become more widespread. With existing car-sharing programs, the marginal cost of driving is higher than for individual ownership and this encourages the use of less expensive modes like transit when the car isn’t needed. The ability to select the most trip-appropriate vehicle and mode is likely to lead to greater diversity of vehicle types and modes at different costs.

  21. bradtem.myopenid.com January 3, 2013 at 10:46 pm #

    Generally the term “driverless car” is pretty widely disliked. These cars have something driving them — it’s just not a human. The world has not settled on the popular term yet. Terms like self-driving car, autonomous vehicle, robocar are seen.
    In the USA at least, you can get by observing the traffic laws by and large. Double parking is not a problem because these cars don’t ever not have a driver in them so they are never parked.
    And yes, the future won’t be quite like what we paint, I agree. But one thing I know is it won’t be like how things are today, and nor will it be like a simple linear extrapolation of today, which is what most transportation planners assume, though perhaps not our gracious host.
    @Ben seems to feel that just because some past predictions of big changes for the future were wrong, they are all wrong. That’s entirely wrong, I suspect.

  22. Miles Bader January 3, 2013 at 11:19 pm #

    @brad
    Nobody cares about the exact term; people know what you mean.
    Current transportation planners have the luxury of lots, and lots, and lots, of existing examples that provide some kind of solid basis for investments.
    Driverless car proponents on the other hand have… well, not much besides faith.
    The future will be different, but if there’s anything all the failed predictions of the last century should show, it’s that the future will be (1) in many ways, much the same as today (prediction of radically different societies rarely pan out), (2) in a few ways, crazily different, and most importantly (3) the crazy differences won’t be the ones we predict.
    So although current long-term transportation planning obviously has some element of risk, there’s little we can do about it, because it’s inherent in planning fo the future—and in any case, it’s a lot less risky than people betting on pie-in-the-sky utopian solutions (because by presuming specific radical changes, they’re less likely to pan out).

  23. Lee January 4, 2013 at 12:38 am #

    Well, the plans that are based around automated cars have the advantage that they don’t need anything new to be built. Most of what they do call to be built (Freeways and rapid transit) will be useful in nearly all foreseeable futures.

  24. Theo January 4, 2013 at 1:22 am #

    Just a quick fact check:
    Google’s automated cars have been driving in mixed traffic on local roads, arterials, highways, and freeways. It’s navigated Lombard Street in traffic with throngs of tourists. Google has driven over 300,000 miles accident-free with occasional human intervention, which is far more than the average human driver can claim.
    Not to be a huge fanboy here, but I think we’re underestimating the leaps the technology has made in the past few years.
    It’s important to separate the facts from fantasy, however. The reality of the intermediate-scenario situation is considerably less exciting than most automated car proponents tend to proffer:
    -speed limits will remain the same, and automated cars will drive exactly at the posted speed limit, which is usually 5 mph under the average traffic speed for a particular road. So automated cars will, in fact, be the slowpokes.
    -as automated driving technology advances, the statistical probability of accidents will decrease, and therefore insurance on automated cars will decrease. (this is of course assuming that the current car-ownership model still persists.) This will accelerate the mode-share switch.
    -conversely, insurance on manually-driven cars will increase dramatically, especially as the remaining drivers will carry significantly higher statistical liabilities than automated cars. There will always be people who drive “manually”, such as car enthusiasts and the rich, but the majority of people will likely switch over to automatic cars for the cost.
    -automated buses will probably arrive in the first wave of automated car acceptance. BRT’s are the most likely candidate for automatic conversion.
    -transit still has the geometric advantage over automated cars. Automated cars will probably first serve a role as a feeder system to stations in low-density areas (like taxis), and as an expensive alternative to local bus service in mid- to high-density areas.
    -if used properly in planning, the adoption of automated cars may allow us to narrow existing roads to reflect a more balanced use of transportation modes.
    – if used improperly, the adoption of automated cars may create greater sprawl and worse congestion for us all.

  25. John Smith January 4, 2013 at 2:12 am #

    Whilst I’ve found this discussion to be quite fascinating, I think we’re all missing the point here. Surely the future has to be ‘transit-less’ towns; places where we live, work and are educated all inside a walkable distance, without the need for any form of mechanised transport whatsoever? A total reversal of ‘urban sprawl’ into individual, sustainable communities. A world where the population gravitates towards gentler climates and people give up the attempt to live in deserts, or on ice fields or geological faults; a world less dependent on burning fossil fuels just to move about or keep warm.
    I’ll wager that our current western lifestyles will become nothing more than a footnote in history… but I don’t think anyone from current generations will be around to collect the bet!

  26. Eric January 4, 2013 at 4:31 am #

    “Theo”‘s post is right on.

  27. Eric January 4, 2013 at 4:36 am #

    John Smith: You can go back to living in a cave if you want. But I enjoy using products developed and distributed with the help of mechanized transport. Health care from Harvard Medical School, electronics from Silicon Valley, orange juice from Florida. It is possible to use nature without abusing it.

  28. Neil January 4, 2013 at 9:00 am #

    I driverless cars come to be, I could very much see them replacing taxis, and coverage-oriented transit service.
    The large-taxi-company business model is not noticeably different whether there’s a driver in the car or not. The owner-operator model is, and this is a possible way that new technology could outcompete the old technology.
    But the driver is only a small part of the cost of running a vehicle. Capital costs and fuel costs are significant. Cutting out the driver from the taxi system will reduce fares, perhaps shifting a few people at the margins from transit to taxis. But they’re not likely to make them cost competitive to transit, nor is there a social-policy goal where it would make sense to subsidise these auto-taxis to make them competitive, except in areas where they can replace low-ridership transit at a lower cost.

  29. Neil January 4, 2013 at 10:14 am #

    @zefwagner “The big danger will be that even when all trains and buses are able to be driverless, they won’t be due to powerful unions protecting drivers’ jobs.”
    What are these “powerful unions” going to do to prevent agencies from upgrading their fleets? Go on strike?
    I think you don’t understand where union power comes from. It’s the ability of groups of workers to collectively withhold their services, and exact concessions as a result of the problems caused by them withholding services. Withholding services to prevent your employer from replacing you (either with technology or outsourcing) isn’t an effective way to convince them to keep you on.

  30. Daniel Howard January 4, 2013 at 11:15 am #

    Wow. A lot of opinions here.
    The current Google Car can operate on city streets autonomously, but it needs someone doing the backend work of getting all the streets mapped out perfectly, figuring out exactly where the lanes are. Then in order to do a truly autonomous taxi service, you’ll want a two-way video linkup for the dispatcher to pilot the car if it gets stuck in some situation like the fire department blocking the street, or to monitor security.
    For that reason, the current livery model works really well: a small, local company will service its fleet and its IT needs. The biggest expense, the driver, will be eliminated. This will serve an evolutionary role of a taxi service within a limited service area. This will be mostly shopping trips for car-less people, and “last mile” services to transit connection points, like Taxis serve now. The evolution comes with lower cost: short-haul, off-peak commuter needs, more “last mile” transit service where an autotaxi will be faster and more convenient than the local bus service, but also cheap.
    What happens next? “Roaming” agreements among carriers sharing a common technology platform. The service areas of the autotaxi companies grow larger: your local autotaxi can drop you off on a shopping trip to a regional big-box store two towns over and the local autotaxi there can bring you back cheap. Expanded mobility, less reliance on transit.
    This doesn’t mean the end of transit. Individual automobiles still require more energy and infrastructure to operate. The autotaxi will dominate short trips, but especially at peak demand, we will need to rely on higher-capacity transit backbones.
    The biggest driver of the need for peak-period transit handoff is the capacity limitations of the autotaxi carriers. You simply can not carry everyone, but you want to be a part of the picture. So, yeah, the service gets you from your house to the transit hub, maybe work out relationships with local transit agencies so thaty “last mile” can be served by auto-taxi as a part of the transit fare itself.
    The other limitation is for longer-range travel, even a fully autonomous rubber-on-pavement highway system will not be able to match the speed of rail-based or air travel. The autotaxi might drive you fifty miles to the high-speed train station, but then you’ll board the bullet train for LA which will be faster and charge a lower fare.
    Anyway, the roaming evolution will mean that we go from local taxi service to regional airport shuttle service, and this will be great for those who live some distance from a long-haul transportation hub who want to make it to/from the airport, &c.
    I think autonomous cars are a very reasonable evolution on human-piloted cars, which were a very reasonable evolution on horse-drawn carriages. In the twentieth century we evolved from horses to humans, and in the twenty-first we will evolve even more seamlessly from human to computer.
    Our streets didn’t change much from the carriage to the automobile era. They’re wider and too dangerous for people to walk in. I doubt the streets will change much in the autonomous era, except they’ll narrow again and it will be safe to walk, bike, and play in them again.

  31. Lee January 4, 2013 at 11:16 am #

    @Neil – mechanics and drivers are often in the same union, so unless if you have a plan for robot maintainers as well, union power is still there.

  32. Robert Wightman January 4, 2013 at 11:23 am #

    Rail transit vehicles have basically one degree of freedom, they travel in a linear fashion. They cannot suddenly change into your lane or make turn across your front, though I will admit that street cars in mixed traffic can make a right turn from the left lane. Automated rail traffic always operates on private right of way so it does not need to look out for stray cars or pedestrians.
    Automobiles have two degrees of freedom as they can move from side to side as well as go along a road. This makes control at least one order of magnitude more complex. Along with this we have pedestrians, stray dogs, and what ever else might enter the roadway.
    Next come the two unanswered problems:
    1) Storage of unused equipment. They take up space so do you send them empty back to the yard or do you build duplicate storage facilities down town and up town. Of course you could just let them keep running around empty.
    2) Energy, from where do these vehicles get their energy? Batteries do not have a high enough energy density to be useful and re fuelling with a petroleum product would be time consuming. Don’t get me started on Hydrogen and fuel cells. I suppose they could be like the “Dodge ’em” cars at amusement parks. The road would be metal and at ground potential and we could have a wire mesh over head to provide electrical energy.
    I am waiting for the personal automated helicopter to get you to work. That would introduce a third degree of freedom.

  33. Frosty January 4, 2013 at 11:32 am #

    I haven’t seen mention of bugs in the software, hackers, server farms going down for a multitude of reasons, snowstorms and other weather events, solar storms, etc. The bigger your grid, the harder it falls.
    This discussion reminds me of the Futurama Exhibit by Norman Bel Geddes at the 1939 New York World’s Fair. Everything was clean, green and pristine unlike the World of Today.
    A more likely scenario is that we will be so caught up in coping with global climate change that our lifestyle will revert to the 1920’s with modern conveniences.
    The future is predictable only by hindsight.

  34. EngineerScotty January 4, 2013 at 12:35 pm #

    I think you don’t understand where union power comes from. It’s the ability of groups of workers to collectively withhold their services, and exact concessions as a result of the problems caused by them withholding services. Withholding services to prevent your employer from replacing you (either with technology or outsourcing) isn’t an effective way to convince them to keep you on.
    Union power can come from several places.
    In some cities (NYC and SF come to mind), unions wield considerable political power, in addition to the power they possess at the bargaining table. In these places, management that tries to bargain hard may find itself out on its ear.
    In other places, the government goes out of its way to spite organized labor (both public and private sector) in any way it can. (As Alon Levy has noted, the places where this occurs tend to be transit-hostile for other reasons).
    In still others, certain unions have more clout than others. Here in Portland, public officials eagerly suck up to construction unions, but tend to ignore transit unions. This partially explains the large number of expensive capital projects in the pipe (not just transit), at a time when TriMet has been cutting service hours, and putting the screws to the Amalgamated Transit Union. (Of course, two decades ago, the ATU did have considerable clout, and managed to win for themselves quite generous pay and benefits, which are now slowly being withdrawn as TriMet has to deal with reduced operating budgets; and part of the reason for the poor operating margins has been–again–a focus on capital projects).

  35. bradtem.myopenid.com January 4, 2013 at 3:16 pm #

    Lots to note here:
    Miles Bader: There’s tons of data on car use. Transit use is just a small fraction of it. What we know from that data, among many things, is that in today’s cities, most trips are no longer in the old “in and out of CBD” style, and transit has trouble serving them.
    Actually, most of our dramatic new trends (like the internet and many of its consequences) were predicted and known well in advance by people in the know. Not the general public, that’s all. The predictions of impossiblity were also quite unreliable, especially by the public but sometimes by experts.
    No, the cars with people in them will go the speed people set. Up to the people to decide if they wish to obey the law, not the machines.
    Insurance for manual drivers should not go up much. Insurance shares the cost of accidents over all drivers. Unless accidents by manual drivers goes up, their insurance doesn’t. However, if the robots are safer their insurance goes down. Safer cars are a win for everybody.
    All-walking towns are interesting (and enabled a bit by robocars) but the truth is you can only force things so much. People will get what they want. If they have children, they seem to want large lots, quiet streets, safe trips to school etc. They give up a lot to get those things.
    The driver is not a small part of the cost of running a Taxi. In Manhattan, it’s 57% of the cost.
    No, cars are actually more energy-efficient than off-peak transit, and modern efficient CNG or electric cars are a LOT more efficient than off-peak transit. Over time, I expect people to move off of off-peak transit into robocars for the convenience, and the planet will be better for it once those services shut down.
    @Frosty: I don’t mean to imply that software reliability is trivial, but your current cars, airplanes, elevators and much more are all running critical safety functions through surprisingly complex software. It’s not easy, but it’s not impossible.

  36. Miles Bader January 4, 2013 at 4:12 pm #

    @brad
    You miss my point. There’s tons of data on both car usage and transit usage. However, you can only use current car-use as an indicator to the extent that driverless cars are the same as current human-controlled cars—and it’s already abundantly clear that human-controlled vehicles are a complete and utter disaster, having destroyed much of the American landscape, at great cost.
    So you have a choice:

    1. Assume that driverless cars will be “mostly the same” as current automobiles; if this is the case, then it’s clear they are a bad thing, and should not be encouraged or supported. Cars have done enough damage to America (and the world, though a bit less), and it’s time to stop the insanity.
    2. Assume that they’re a radical departure (which of course could be possibly for good … or possibly bad); in this case, existing data is not very useful, and there are other, much better understood, ways of curing America’s destroyed landscape.

    That doesn’t mean such “radically different” driverless cars are inherently bad, just that they should not be used as an excuse to stop investments in other more mature methods.
    The real issue is the maroons who say things like “why should we build this subway / bike path / … driverless cars are coming (Any Day Now)… they’ll solve everything!!1!”

    Actually, most of our dramatic new trends (like the internet and many of its consequences) were predicted and known well in advance by people in the know.

    Sure they were… (“people in the know”…heh)

  37. Miles Bader January 4, 2013 at 4:24 pm #

    p.s., I think my point should already be clear, but add to my list:
    3. Don’t assume. In this case, of course, we should continue to use known methods to address current problems, and that means discouraging car usage, and building more walkable, more transit-friendly cities. Those solutions actually work, and delaying just increases the pain.

  38. bradtem.myopenid.com January 4, 2013 at 4:35 pm #

    Well, they do change the landscape — for both cars and transit. But they’ll start looking very much like cars and taxis, and as they grow have a grander effect. Let’s face it, outside of Manhattan and a few other places, transit ridership is quite minor compared to car use, and so the effect on car use is probably much grander in its effect on cities, but both are interesting to study.
    The data we have today covers who lives where and what trips they take, and often what modes they use. Initially, this does not change much. Eventually people may change where they live based on new transportation options. So I’m not sure why transit analysis has any particular data advantages.
    It is my contention, for example, that Manhattan’s taxi fleet is vital to it having the lowest car ownership by far in the USA. Manhattan has the biggest transit system too, and many ride it, but the thing that lets you give up the car is that you can ride the transit where transit goes, but you can get anywhere else in the cab. It’s the “everywhere” that’s important.
    Outside Manhattan people look at the world and say, “If I don’t have a car, there’s a lot of places it’s really hard to go. Thus I will spend a huge wad of cash on a car. Ah, now I have a car, and having spent that money, I would be silly not to use it and take transit.”
    Robocars — and taxis — break this cycle. And so yes, they are important if you are debating should I put in a subway or LRT. (Not sure if they have a lot of bearing on bike paths, other than the ability to go one-way by car and then one-way by bike makes the bike path much more appealing for a commute, if you have something — robots or people as in Velib — to redistribute the bikes.)
    If I were trying to plan future transit lines I would focus on where and when the lines are most efficient. I would plan not to run them at off-peak times. I would consider if my ROW can carry other automated vehicles when not in use. I would expect people to be much less tolerant of doing transfers or taking trips that go way off the direct route. I would go with BRT over LRT if I was doing either at all. I would expect it to be easier for people to get to and from stations from further distances. I would design platforms for easy mode-switch from train to robocar. I would (if doing BRT) have different sizes of vehicle and dynamically allocate the right sized vehicle to the load.
    And many other things. At least if I don’t want my plan to seem silly in the late 2020s.
    I note you say we must discourage car usage. Why, if the cars use less energy than the transit vehicles, is that a must? Don’t we want to encourage the most energy-efficient forms of transport? As I describe in robocars.com/rod.html we have the opportunity to improve walkability a lot as the robocars gradually reduce parking lot demand and we naturally infill and push for walkability in the urbs.
    I am not sure what you mean by “heh.” I’m serious. I was there. I was one of them. We had mailing lists in the 70s on the arpanet where we talked a lot about how the world was going to change and get connected. And it happened. Yes, the general public was shocked in 1994 when the internet “exploded” but for many people that was not a giant surprise. (Or rather, what was a surprise was how surprised people were.)

  39. Scott January 4, 2013 at 5:03 pm #

    Neil21 writes, “@Scott is scared by driverless cars, but presumably not by human-driven cars, skytrains or autopilot. Not a credible forecaster.”
    Oo, nasty and egotistical! And rather odd, since my impression is that overall we agree with each other.
    Part of my point, perhaps not well expressed (though on the other hand my impression is you weren’t listening, since you clearly did not read my second paragraph, or if you did you ignored it) was that my fears are – as you point out – based in my openly recognized and automatic human biases and what I am used to. Indeed, because I am used to some things that struck my grandparents as ridiculous and useless, they do not bother me at all, while my grandchildren will – I suspect – consider self-driving cars just part of the landscape. [Reference here Heinlein’s “Columbus Was a Dope.”] However, my fear does not control even all my decisions, and surely – even combined with others’ fears – does not control everything that happens in society and tehcnology over long periods of time! Nor can any of us see all the outcomes or steps that will happen over those long periods.

  40. Miles Bader January 4, 2013 at 6:01 pm #

    @brad
    Manhattan’s public transit system is great for a U.S. city. It could be much better though, and is underfunded and underdeveloped (it’s “great” for the U.S. in part because the rest of the U.S. is so awful). Even Manhattan, after all, was a target in the “remake the landscape for automobiles” movement, no matter the cost.
    We certainly need some individual motorized transport, and yeah, taxis are a good way of providing that (so probably the most realistic way of thinking about driverless cars is as cheaper taxis). But it’s only one part of the transport system, and especially in a place like Manhattan, a system based only on individual vehicles isn’t viable due to the amount of space needed (low-speed individual transport like bicycles are better, but not so good for longer distances). A good transit system mixes modes efficiently, and one of the biggest problems with America’s car-obsession of the last 50 years is the nutty assumption that cars were the magical technology that could adequately assume all transit roles (and the insane things that were done to accommodate that assumption).
    The problem with cars is not the amount of energy usage (though that certainly sucks too), it’s the (1) the insane amount of space they consume, for a given level of transport functionality, and (2) the dysfunctional urban landscape that has resulted from car-oriented development.
    (1) of course has a strong effect on (2), and some of (2) is incidental (people had many bad ideas at the same time). Driverless cars will be able to address (1) to some degree (less parking, better road space utilization), but there are limits, and even getting close to them requires much more radical changes (e.g., no human drivers allowed, special roadways [which are likely to be more expensive and/or have a more problematic effect on the local-scale urban landscape])—and as soon as radical changes are required, you’re well into “be careful what you wish for” territory.
    Again, my point is not that driverless cars are inherently bad, or won’t be a big deal in the future. Rather, I’m just saying that it’s a very bad idea to bet the farm on the shaky theory that “driverless cars are the solution.”
    Yes I’m sure you’re proud that you saw the potential the internet (me too, btw…). But lots and lots and lots of things are predicted, and relatively few pan out. In those cases where the predictions do pan out, it’s natural to think “I knew it!”—but having been right occasionally in the past obviously isn’t by itself a particularly good indicator that a given future prediction will be true.

  41. Matt L January 4, 2013 at 9:54 pm #

    I looked at some of these issue a few months ago. While the technology sounds good and I have no doubt they can get it to work properly, I simply don’t think the proponents have thought it through that much.
    1. There will always be a small element that want to use the old fashion method and we will still need to cater to them.
    2. In NZ at least, based on current vehicle numbers and new vehicle registrations, it would take 20+ years to replace the existing vehicle fleet and that is even if they came on market tomorrow and every single car sold had the technology in it.
    3. While driverless cars could use less roadspace, the need to cater for existing vehicles will still limit the capacity of roads, there will likely be some improvement over the current capacity but not a great deal. When there are a lot of these cars running around, they should all be able to run closer together but for pedestrians that means you would only ever get a chance to cross roads at designated places as otherwise you will be faced with a solid wall of vehicles.
    4. There would still need to a lot of expensive parking infrastructure. Some driverless car promoters suggest the cars would instead be able to drive themselves home and park there but then you are doubling the number of km’s each vehicle has to travel. It also removes the advantage that cars are said to have of spontaneity, say you live 20 minutes away and need to leave work early, you have to wait for your car to get there first. The other option would be for them to effectively replace taxis but in most cities, that is a massive shift culture shift.
    5. On the positive side, these cars will be much safer as the technology should allow for them to detect and respond to the environment around them better, plus they won’t be speeding. There is a massive elephant in the room that the proponents have yet to acknowledge though. In a slight contradiction to point 4, that safety feature will mean that anyone or thing could step out on to a road and the entire line of traffic will grind to a halt simultaneously. Want a more pedestrian friendly and less car dominated town centre then just keep stepping out onto the road and the computers will quickly communicate with each other that something is wrong and the rest of the traffic will find a different route to avoid the area. They could be the ultimate tool to reclaiming cities from cars and pedestrians can laugh while watching car riders fume not be able to do anything about it.
    http://transportblog.co.nz/2012/10/28/are-driverless-cars-the-future/

  42. bradtem.myopenid.com January 4, 2013 at 10:50 pm #

    Matt L:
    1) Sure, they will still be around, what of it?
    2) As the important part of a car becomes a computer, this math changes. Cars get used much more (they are shared) and reach their end of life much sooner. But they’re disruptively better, so much better that people junk old cars for scrap sooner — akin to, though not quite to the extent that we junk old perfectly working computers and phones.
    But yes, it still takes time, even with this.
    3) Yes, but a lot still happens in the meantime while the old vehicles are out there. Some improvements are more ITS than robocars, and apply to an old car with a smart phone in it. Yes, at rush hour peds may not want to jaywalk — it was always illegal.
    4) They don’t have to go home. Even without the fact that robocars could valet park at 2.5x or more density we actually have an oversupply of parking because people demand parking very close to where they are going. Robocars won’t – half a mile away is just fine, so you balance over all the lots. And the taxis don’t park at all, the go off to serve other people. And the robocars never actually park, they stand, and they can do that in front of non-busy driveways, fire hydrants, you name it.
    5. Yes, but the ped who does that is photographed by the people in the car that came to a stop. These cars all have cameras pointing forward. I don’t recommend you doing it a lot, not with the face recognition that’s going to be out there in a few years. You’ll find a fat jaywalking ticket in the mail. Not sure I like that but it’s probably how it will be — the people you you walked in front of will be very annoyed.

  43. Matt L January 5, 2013 at 12:13 am #

    Brad
    1. it means we still have to build out roads and cities around the fact there are non computer controlled vehicles so no squeezing of road space, no massive capacity changes. In fact even with driverless cars we will probably still want cars of the same dimension for the same reason we don’t have two seater cars with the driver in front and the passenger behind.
    2. Yes they may result in a faster uptake but I can only see that happening when the cost is so low that it becomes a standard feature, that is a long way off as the systems systems needed are still 10’s of thousands each.
    3. The point is by driving cars closer together we aren’t magically going to double the capacity of roads as there will still be limitations at places like intersections. Even if you did double the capacity of roads, that doesn’t necessarily mean that there won’t be congestion or that transit isn’t needed. Also where I live in New Zealand it isn’t illegal to jaywalk unless you are within 20m of a signalised crossing. Even then the police would never even consider saying anything to you. Making jaywalking illegal is one of the stupidest ideas ever.
    4. Where they go after dropping you off depends on how the system is implemented, if it is private owners then they need to go home or to a car parking building and I doubt you could get 2.5 times more density out of those. If it is a fleet of robo taxis then that fleet would need to be massive and as such the owner would likely need to set prices quite high to get a decent enough return on them considering most would be only for peak demand. That will put people off using them compared to other modes or old style cars.
    5. Again it isn’t illegal where I live to cross the road and it also wouldn’t have to be a person but could be any kind of obstacle like some balls. Even today GPS systems that can change suggested routes based on congestion, with automated cars/taxi’s that will happen much faster. Also are you suggesting that every private vehicle will feed facial recognition data back to central authorities, good luck with that.

  44. Hibai Unzueta January 5, 2013 at 3:22 am #

    Interesting conversation. Thank you all.
    Although the evolution metaphor is useful to measure the difficulties of implementing automated cars, commenter Lee has pointed a credible gradual evolution scenario, where cars incorporate the automation before they are left to do all the driving. Therefore we will probably start to see them among us in the future.
    But the real question here is: what benefits will this automation really provide? And, would that produce anything capable of replacing transit?
    My opinion is no: Although we already have automated trains and the Google Car driving around, the complexity they handle does not match the promised compact circulation scenario. And if they will not be able in the foreseeable future to provide a more compact circulation, a better use of the available street space, I have to ask: What is then — apart from improved safety or less operating costs in the case of taxis — the big advantage of these vehicles?
    And the answer, although it might not seem obvious due to our high ideological bias is that, they are private! What is the most important part of PRT? Obviously the P part in it!
    Don’t take me wrong: I do not by any means advocate for the mass use of private means and I envision transportation as something shared both by geometrical reasons and social ones. But if I am to tell why the automated car future is so promising for so many people, the main reason I can think of is that it is still a private enviroment.
    Well organised, high frequency transit is like a conveyor belt or a moving street, to picture it more appealingly. The conveyor belts are placed to match the common denominator of the movement of hundreds of people. And it is that way that they can achieve efficient energy and street space use.
    But private vehicles, either automatic or manned, either shared or not shared, always hit the limitation in efficiency of not being part of any ‘conveyor belt’ — which is why they can provide point to point transportation.
    That being said, automation may come to private vehicles, but geometry will foreseeable prevent that from being a revolution in the way we move. We will still need automated or manned public transit based ‘conveyor belts’.

  45. SW77 January 5, 2013 at 11:08 am #

    http://www.reddit.com/r/SelfDrivingCars/ has a timeline in the sidebar that predicts when major milestones in the adoption of self-driving cars might occur

  46. Jim January 5, 2013 at 5:00 pm #

    Driverless buses. The reason I can see that they won’t last is that too often buses are empty. Sure, if you get a few lines that are always full they might last, but most aren’t.
    The problem with buses is that the owners have to plan the route, and hope there are riders that will ride it. Riders won’t ride a particular bus until they known about it, which generally means it’s going to have to pre-exist before they get riders.
    Sure, take existing lines and make them driverless. But then you have the other problem, that riders still have to meet the bus to ride it.
    With driverless cabs, theoretically the cost will be able to be low enough that they can ride it for what it costs them to ride the buses today, and make their trip much faster than trying to take buses.
    Because that’s the other problem with public transport. You have to schedule your life around them. You have to try to make the bus or train. Miss it and you’ve got to wait for the next one. Then it’s also slower because it’s not necessarily going the fastest route to your stop. I say your stop because your stop isn’t necessarily your destination. In fact it probably isn’t.
    With a jonnycab, you order the cab to pick you up at your start location, then go straight to your destination. No 20 minute wandering route to pick up other people who might be at stops. Just straight to your destination.

  47. Matthew Newton January 5, 2013 at 8:56 pm #

    Good discussion all.
    While the initial post has been responded to effectively, there’s two key points which have been completely ignored and they are probably THE two key points which ensure the success of autonomous vehicle tech.
    1. All dominant systems harness human nature and don’t fight against it. That’s why driverless cars are such an attractive technology. Public transport advocates have spent so many years correctly arguing for increased implementation of transit but have met years of frustration for the one simple reason : they are fighting human nature.
    Human nature dictates that people want on demand, point-to-point and private transit. The closer any technology comes to this, the more successful it is likely to be and THIS is the key reason why driverless cars will succeed.
    If you’re a transit consultant, you might dislike this but you’ll have a much better impact on the world if you accept them into your calculations and consult with governments in order to plan and prepare.
    2. Another thing that is forgotten by most is the incredible potential in terms of Logistics.
    The business of logistics alone determines that self-driving cars MUST be a success. Look at the huge rise in package shipping because of the internet or the current trends towards same-day or next-day shipping from online purchasing. Companies like Walmart who spend billions on logistics will quite happily use driverless trucks if it reduces their costs. Quite simply, there’s no misty-eyed sentiment in B2B land about the magic of acceleration. If a business case can be made, it will be done. Just look at Rio Tinto, who have developed autonomous tech for driving trucks around mines. That is but one step on a huge journey.

  48. EN57 January 6, 2013 at 8:05 am #

    Matthew Newton. No, I don’t think it’s human nature to want to be handled like a package in a freight distribution system. But even these systems use higher-capacity vehicles (eg semi-trailers, 747s etc) where required to move a lot of packages along certain corridors more efficiently.

  49. John Niles January 7, 2013 at 12:57 am #

    Thanks to all for a very substantive discussion on the prospects for wide adoption of self-driving cars.
    From reading through all the commentary that was volunteered above, it seems to me that there is plenty of information in the comments (including the hot links) to challenge the posted concern of Jarrett Walker, who wrote, “I will begin to take driverless cars seriously when I see credible narratives about all the intermediate states of their evolution, and how each will be an improvement that is both technically and culturally embraced.”
    Mr. Walker’s requirement for seriousness is a challenge to fulfill, but based on all that’s here, I’m feeling better already that there is a potential for evolution in stages toward a mix of robo-cabs and privately-owned (small, clean, quiet) vehicles that move smoothly in congested traffic and don’t hit each other so often. Plus, don’t hit pedestrians and cyclists so often. And won’t public transit agencies — faced with rising costs per boarding across America — take advantage of automation also?
    I hope Mr. Walker will now chime in with a new assessment of motor vehicle automation after all the human input so generously offered here at humantransit.org.
    Of additional interest, I’m on a research/design team led by the Connected Vehicle Proving Center at the University of Michigan, funded by the Graham Environmental Sustainability Institute at the same University, assessing how the evolution of road vehicle automation, coupled with oil-free propulsion and ubiquitous machine-to-machine wireless communication, could/should/might co-evolve with the livability scenarios stemming from the land use and mode shift designs encouraged by Federal, state, and local governments.
    We want to find an evolutionary path supported by public policy that supports sustainability broadly defined … environmental, efficient, effective, economical, equitable.
    We think so far that smaller, quieter cars with automation can be compatible with evolution toward more walkability and bicycling in urban areas.
    And what are the likely objections and problems to arise? See the comments above for clues, noted.
    Working through the rest of 2013 toward a published technology and policy road map, my team is using the Seattle-Tacoma-Bellevue-Everett area of Washington State as a geographic focus and case study because we know this territory and its leaders, with whom we discuss our work.
    Here’s a related fun fact — the latest official MPO Metropolitan Transportation Plan for the Seattle region approved in 2010 forecasts the private vehicle market share of 2040 motorized trips at 95%, transit at 5%, and the number of registered motor vehicles per capita as roughly what it is today. How likely is that?
    There is some early visibility on our “integrated assessment” work at http://www.aboutcates.org . I am doing this work as part of CATES, the Center of Advanced Transportation and Energy Solutions, a new think-and-do tank in Seattle.
    Jarrett Walker’s challenge, and the comments in reaction, are great input to my team’s work. Thank you.

  50. Jarrett January 7, 2013 at 6:00 pm #

    Thanks for all the comments! I will respond soon, though it may take a few days … Nag me if I let it slip … Jarrett

  51. Henry January 7, 2013 at 6:23 pm #

    A couple of things, just my two cents.
    There are several issues with driverless cars that I would like to raise.
    1. Liability
    If a car hits and kills a person, or crashes and destroys property, whose fault is it? The programmer’s? The government’s? The manufacturer’s? This would lead to very messy court battles.
    2. Reliability
    Driverless cars are great, sure, but what happens if there is total network failure in an area, like the 2003 blackout or Hurricane Sandy? In this imagined future, driver’s licenses will probably be unnecessary, so a significant portion of the populace would be stranded in their homes or would get themselves hurt trying to figure out how to drive a car.
    3. The economics
    Sure, the economics of an individual car would be improved drastically. But with shared cars, who would maintain an entire fleet? You won’t be able to have a single company in control of it, because that would lead to a cushy monopoly. You wouldn’t have the government to do it – it is not the government’s perogative to ensure that everyone has a car, a house, and a spouse with two kids. You’d also need a place to store these cars – without off-street parking (which should not be dominated by corporations, in any case), they’d need to maintain large, unsightly depots with shops, similar to bus and train depots. Hmmm…
    4. Interactioss with other modes
    Yes, everyone has talked about Google and their fantastical cars, and they’ve navigated San Francisco, but California is not exactly an extremely difficult place to navigate. You have places like the streets of Manhattan, where pretty much everyone crosses wherever they damn well feel like. You have shared spaces, where all modes interact without barriers. I’m not even mentioning the chaotic roads of the developing world here. Can these driverless cars navigate around these obstacles?
    (Before someone here mentions that everyone who jaywalks should be fined, that is absolutely ridiculous and stupid. Jaywalking was not a legal concept until the 20th century, when automobile manufacturers imposed their own traffic rules on previously shared roads. The purpose of a city street should not be to prioritize car traffic, but to benefit the community, and if they choose to jaywalk, then so be it. In addition, there are many places in the United States where roads do not have sidewalks or crossings in residential neighborhoods, and it is stupid to suggest they should get a driverless car to make a U-turn and move them 40 ft across the street.)
    5. Navigation
    I presume most cars will either be using GPS or some sort of technology based on map databases. The problem with this is that there are a LOT of places on the world’s mapping services that are either unlisted or in the wrong place. For instance, my house in the borough of Queens, New York, is listed on every mapping service as being a mile away from where it actually is. I would not enjoy being dropped off by a driverless car a mile away from my house. There are also things like road closures and detours that aren’t currently integrated into mapping services.
    Sure, the government could maintain a map database with every home and business, but this has issues as well: many businesses are defunct or shell companies, and it is a massive violation of privacy to have every single home in a database.
    6. Everyone is assuming that the bus of the future is your standard diesel bus.
    Buses are not very energy efficient, but they take up less space than an equivalent amount of cars. In addition, buses have large amounts of room to store battery packs in, and their frequent stops allow them to take advantage of regenerative braking. Regenerative braking combined with battery packs would greatly increase the efficiency of buses.
    6. This only encourages a car-centric lifestyle.
    There are people here who don’t see greater car use as a bad thing. I’d like to point out that the United States, the Western country with the most car-centric lifestyle, also disproportionately suffers from obesity, diabetes, asthma, and coronary diseases. This is because Americans don’t walk nearly as much as their counterparts in Europe and Japan – all they need to do is take a car everywhere. Encouraging this sort of lifestyle would be bad for public health.
    Sure, you can say the “free market” demands such things, but how much of this is just the free market? The US spends more on roads in a year than it has spent on Amtrak in its entire lifetime. Combine this with zoning that forces developers to build more parking than they need or want and restrictive single-use zoning, and you get a market heavily distorted in favor of car use.
    To put it neatly, there are too many ifs for me to be comfortable with driverless cars, and we need new thinking about transportation instead of the singleminded “If we improve cars, we improve the lives of all Americans!” Just my two cents.

  52. Jack Horner January 8, 2013 at 1:16 am #

    1. ‘Smaller cars’ and ‘driverless cars’ are different issues. We could be building cars that fit three abreast in two lanes already. Why aren’t we, and how is the arrival of driverless cars going to change that?
    2. With or without drivers, a big city with 95% of trips by car must still have a lot a of traffic congestion and a lot of roads and parking lots, with all the urban ugliness that that implies.
    3. So a driverless car, to economise on parking lots, could drive itself home after dropping you off. What effect would that have on the general traffic and congestion level?
    4. No doubt driverless cars will increase the capacity of existing roads to some extent by taking the human element out of reaction time. By what extent? I think it would be courageous to expect this marginal efficiency gain to solve the problems at 2 & 3 above.
    5. Increasing the capacity of existing roads (that is, a denser flow of traffic) will be hostile to the amenity of pedestrians. The fact that a driverless car might be able to stop quicker, if you are so bold as to step out in front of it, is not a solution.
    We are back to the problem of basic geometry: with or without drivers, 1.2 person private vehicles take up an *enormous* amount of urban space compared with other forms of transport.
    If you’re happy to live in car-dependent sprawlsville that has been designed to provide this space, this may not be a problem for you.
    If you value some of the amenities of urban life that sprawlsville cannot provide, you still have a problem that driverless cars won’t solve.

  53. EN57 January 8, 2013 at 4:56 am #

    The auto industry has planted its seeds well, ensuring enthusiastic and unquestioning demand for its products well into the next generation.
    I always thought the freedom to control the steering wheel was part of the deal with private cars; but from the above discussions, personal freedom and control don’t seem to be issues. – Just get into a pod, punch in your destination, and your favourite courier company will get you there in the quickest way possible (subject to the demands of other [platinum class priority] users)…
    I support the points Jack Horner has made. There are no solutions to urban traffic congestion with these technologies. Just more people spending more hours, days, years stuck in more traffic. The freedoms of walking and cycling, or the efficiencies of transit don’t come into these discussions. But where does this eagerness to embrace computer controlled private vehicles come from?

  54. Jim January 8, 2013 at 5:38 am #

    “Matthew Newton. No, I don’t think it’s human nature to want to be handled like a package in a freight distribution system. But even these systems use higher-capacity vehicles (eg semi-trailers, 747s etc) where required to move a lot of packages along certain corridors more efficiently.”
    The problem with that is that humans aren’t packages. Packages don’t mind taking 3 days to arrive at their destination.
    In fact, I predict that when driverless cars are the norm, trucks will begin to shrink. The advantage of larger trucks will vanish. The thing that Matthew doesn’t understand is that each truck currently still has to have a driver, which is why transporting larger quantities at once is cheaper. More packages per driver means fewer expensive drivers per package.
    Without drivers, it could easily come down to shipments being the same cost whether in large trucks or small. When that happens, small trucks will win out because it will be faster to send out shipments as they are ready instead of waiting until a larger shipment is ready to go. The only exception will be very large sized items that won’t fit in small trucks.
    But mfrs, stores, and whatnot will demand faster delivery, and mfrs will want to have less product on hand, because the more product on hand means more money tied up waiting to be sold. So if you can ship smaller shipments as they are ready, all the better.

  55. Alexander January 8, 2013 at 12:12 pm #

    The transition is straightforward and will be entirely gradual: Driverless cars completely change the economics of car-sharing; they will make car-sharing much cheaper and eliminate the problem of how to get to the spot where the next available car is.
    More and more people will switch to car-sharing (=driverless taxis) for economic reasons only and thus the share of driverless cars will increase, the share of privately owned cars will decrease.
    People will simply vote with their purse (and it really does not make any sense to spend as much for cars that sit idle 95% of their time as we are used to today….)

  56. Ben January 8, 2013 at 6:47 pm #

    Everyone has brought up interesting points here, but I’m still doubtful that any of this autopilot car stuff will actually ever sell — whether economically or politically. We have enough problems coming our way in the next decade, and I’m sorry, but figuring out how to get everyone into a robot car isn’t one of them.
    As I said previously, we already know how to do efficient travel (just look at the train system in Europe) and here in the U.S. we can’t even muster the money and willpower to make Amtrak even 1/10th as good as what Germany or France has. And I’m supposed to believe the solution is robot cars?
    It’d be wiser to put our collective wealth into a better public transit system, more sensible development, improved biking infrastructure, etc so we all don’t NEED to drive so much. It’s sad how much we spend on roads when I think about what other transit improvements we could have for a fraction of the price.

  57. Matthew Newton January 8, 2013 at 8:25 pm #

    @Ben – the difference is simple. It’s private money, not public money.
    Politicians have been crawling all over this, eager to get their face on it.
    Like I said above, a huge reason that self driving cars are inevitable is down to human nature. Human nature has shown how difficult it is to get any public transport project through, due to the enormous cost, unions, politics, bulldozing people’s houses, etc etc. Just look at the enormous money effort required to just get PART of a railway line built to San Fran, compared to the law to get driverless cars legalised in Calif (the vote went through nearly unanimously in both houses within a matter of months).
    When no public money is required, it is thaaat much easier. Both Larry Page and Sergey Brin are ridiculously excited about this project (both have mentioned it consistently in nearly all media appearances). These guys have billions of dollars and even more importantly, don’t need to ask anyone’s permission to build the cars, so no willpower is required.
    Ben, I used to be in your shoes in terms of advocating public transit. But, a better solution has come along and I’d suggest that if you were to similar levels of research to what most proponents have done, you’d reach the same conclusion! It’s super exciting! Start with Brad Templeton’s website if you’re curious.

  58. Jason January 9, 2013 at 6:10 pm #

    The Australian Government has a discussion paper out and is seeking comment on possible law changes that might be needed to accommodate the intermediate technologies that lead to fully driverless cars. Submissions close January 31.
    http://www.ntc.gov.au/DocView.aspx?DocumentId=2347

  59. Kenji January 9, 2013 at 7:25 pm #

    No single solution is going to solve all the problems. I think it depends on what kind of city you (or most of the population) want to live in.
    I can see driverless taxis being dominant eventually in smaller centres where the buses are probably empty most of the time. It would never be cost effective to run enough buses to be more convenient than a robo-taxi.
    On the other hand, in a large dense city there simply is not enough road space to serve the number of robo-taxis needed when people want to travel. A bus can replace 30-50 cars, which is important when space is at a premium.
    I think the space argument as well as affordability concerns will mean that there will always be a role for mass transit.
    Replacing privately owned cars with shared robo-taxis can definitely reduce the number of cars on the road, but trying to replace a full bus at rush hour (let alone a streetcar or subway) would take more space and make the situation worse.

  60. Riccardo January 10, 2013 at 1:11 am #

    De bono and other innovation gurus have lomg spotted thebproblem with imagined futures.
    First, they are still extrapolations of current phenomena, and miss the transformational changes.
    Humanity is waiting for breakthroughs in several area, including fusion, antimatter, high temp superconductivity, new materials, biology, quantum computing, on it goes. We dont know when, if ever, any of these things will come through. Ten years or one thousand.
    For example, high temp superconductivity might allow the maglev paradise some fans dream about, while new materials might give use flying cars. We might get all the interstellar propulsion we need from fusion, so we don’t need to live on overcrowded Earth. If we could live one thousand years, then a fifty year trip to the nearest stars might not bother us. Ultrafast computing might enable everything to be recreated virtually.
    None of this crap stuff might happen, but if it does, it would changethe world very radically, and invalidate all this extrapolation.
    I don’t like the idea of calling on taxpayer funds to pay for the imagined future either.

  61. Christopher Parker January 10, 2013 at 12:36 pm #

    The evolution that I see:
    1. As mentioned, the technology appears first as an option on luxury cars.
    2. Also as mentioned, the technology appears for freight and fleet users. This includes taxi cabs.
    3. We get new high-speed “lexis lanes” on highways strictly for automated cars – in the manner of hov lanes.
    That’s when we have to ask, ‘what changes?’
    1. The biggest cost and biggest problem for truckers is labor and the driver shortage. Solve it and the competitive balance between rail and trucks changes radically. Result = a lot more trucks.
    2. Taxis become half price. That is NOT the same cost as taking a bus. Not even close. But it does make them more attractive.
    The tipping point is when the taxi business really changes and many more people don’t buy a car and do rely on taxis.
    But even if taxi use doubled or tripled, or by a factor of 10, it’s still a small factor.
    Most people prefer to own a car. Taxi use instead is already cheaper for many people, but is not even considered.
    I also think that a large group of people will need a car and won’t be able to afford the new technology and will keep driving.
    Result: a class difference and safety difference. The haves, will have their time back, their safety back.
    Because a driverless car can go out of sight to park, and because the politically powerful class will no longer care about needing to park close but will care about environments not so marred with parking, I predict more remote parking. The poor will walk. Over time that reinforces the new technology.
    On balance: the new technology will make road use more attractive (including bus), will give us back our time, will make sharing of cars easier (but I don’t see that being a big factor).

  62. R A Fontes January 10, 2013 at 1:47 pm #

    Maybe one way to envision an evolutionary path towards a robocar future would be to look at some developments in the automobile’s past:
    1. electric starters
    2. automatic transmissions
    3. automatic chokes
    4. anti-lock braking systems
    5. electronic stability control
    6. cruise control
    7. adaptive cruise control
    8. lane departure warning
    9. self-parking technology
    10. GPS
    11. signalized intersections
    None of these were developed specifically as an evolutionary step towards robocars, but their general concepts fit in to various computer controlled transportation scenarios.
    On narrower lanes:
    While standard lanes are 12 feet, Oregon’s Department of Transportation does have provisions for 11′ and even 10′ lanes. Robocars’ ability safely to use narrower lanes comes from precise computer control more than vehicle size. Trucks and emergency vehicles will continue to need at least 9′ lanes. One likely source for additional lanes is the elimination of a lot of on-street parking.
    On driverless transit:
    Yes, transit operators should be among the first to adopt robotic technologies and costs will drop. However, once robotic “Zipcars” and autonomous taxis are ubiquitous and competing directly with public transit, taxpayers will wonder if further subsidies are justified.
    Car2Go offers trips at $.35 per minute. A robo-SmartCar would undercut transit for a trip from the supermarket for one passenger, and be a lot more convenient. Imagine a robo-Prius offering a trip for up to five people for that kind of price.

  63. JN_Seattle January 13, 2013 at 1:13 am #

    Now I’m waiting for Jarrett Walker to chime in on self-driving vehicles, perhaps with a partial reconsideration of his opening essay at the top of this present increasingly long page loaded with comments.
    There is some documentation of evolution — both historical and future — toward robo-taxis by Princeton University Professor Alain Kornhauser visible at http://www.cs.princeton.edu/courses/archive/fall12/cos402/readings/kornhauser_slides.pdf
    Further clues to the evolution of self-driving automobility already now underway is suggested in the New York Times of January 12: http://www.nytimes.com/2013/01/12/science/drivers-with-hands-full-get-a-backup-the-car.html

  64. Matthew Newton January 13, 2013 at 5:26 pm #

    Waiting for Jarrett!

  65. Nathanael January 16, 2013 at 5:51 pm #

    The biggest problem with “driverless cars” is that the first time one of them hits someone and kills someone, they’ll be banned.
    Even if they’re safer than regular cars.
    We know this because of the regulations on driverless *trains*. There’s complete fear about having driverless trains with level crossings, so they’re banned The same fear will affect driverless cars.
    There’s some deep psychological bias here: most people are very unhappy about being killed by robots, and are relatively OK with being killed by humans, because there’s someone to blame. This has been true for centuries and it’s not going away.
    Henry’s point #1 — liability — is THE crucial point. Every liability proposal has a fatal flaw which will destroy driverless cars.
    Blame the programmers? Nobody will be willing to admit to being the programmer, and the manufacturer will be blamed. Blame the car manufacturers? Then nobody will manufacture the cars (the liability isn’t worth it).
    Blame the “driver”, who isn’t driving? The unfairness of this will rile and scare people, so nobody will buy one of these “this car goes directly to prison” cars.
    There’s no solution for the liability problem unless you have a fully grade-separated road system. But if you have THAT, you might as well build trains.
    There’s another point closely related to liability:
    For liability reasons, driverless cars will have extremely conservative programming which will maintain safe following distances, and so forth. As a result, since most people drive unsafely and follow too closely, driverless cars will SLOW DOWN TRAFFIC and will REDUCE the capacity of the roads.
    We’ve seen this happen with the installation of safe signalling systems on trains; it reduces capacity compared to the “seat of the pants” driving system.
    So driverless cars will be unpopular and will reduce road capacity. This is not the future.

  66. Nathanael January 16, 2013 at 5:53 pm #

    “Companies like Walmart who spend billions on logistics will quite happily use driverless trucks if it reduces their costs. ”
    Liability. The driverless trucks have to be good enough that Walmart can afford the occasional $100 million payout for its “killer robotrucks”.

  67. Nathanael January 16, 2013 at 5:55 pm #

    “That Google robot car may be able to drive safely on pristine controlled highway conditions, but it would be extremely difficult to make it work in most driving situations. ”
    Like bad weather. It’s had no serious testing in snow, for example.

  68. Nathanael January 16, 2013 at 5:58 pm #

    “But today, ask an urbanist to try to imagine a bus transit experience that could be superior to rail. ”
    It’s impossible, and *we know why it’s impossible*. It’s impossible because rubber tires on asphalt have a *worse ride* than steel on steel. That is the only essential difference. Any other improvement you can make to a bus, you can make to rail, and vice versa.
    At some point, facts come into play. Some technologies are inherently better than others in particular ways.

  69. Nathanael January 16, 2013 at 6:00 pm #

    Whoops, I have to correct myself. Sort of.
    Not being on tracks also means that buses don’t scale, obviously — you can make trains arbitrarily long, but you can’t do so with buses due to fishtailing. That is actually essentially the same cause as the worse-ride-quality problem, but a different manifestation of it.
    Now, do I think we could have “railbuses” and “buses which are almost as good as rail except the ride quality is worse”? Sure.

  70. Joe Busman January 22, 2013 at 12:11 pm #

    The reason nature has not created a wheeled animal is not so much the need for a completely working wheel or nothing. An animal would still benefit from a partial wheel like spokes minus the rim. The problem is that a wheel requires separation where blood or energy cannot be transferred from the body to the wheel part. So far nature can only create “wheel-like” locomotion as in animals that roll up into balls and roll down hills or roll dung into balls to move them more effectively. One possibility is an animal using a wheel shell like a sand dollar, but we also have a problem of axle friction. The animal would have to use its bone or nail as an axle to avoid shearing its skin off, and the wear and tear on the bone axle may not be worth the benefit of faster locomotion. Nature, however, produces something far superior to the wheel for locomotion, and that is the wing for birds and the tail for sea animals.

  71. Matthew Newton January 23, 2013 at 2:04 am #

    Regarding liability, Nathanel you are simply not correct.
    This liability myth has gone on for far too long.
    Google has stated that they are more than happy to accept liability, if that is what it takes to get their software into the marketplace. It was either Larry Page or Sergey Brin who said it. No source. You’re just going to have to take my word for it.
    Mat.

  72. Matthew Newton January 23, 2013 at 2:12 am #

    “http://www.driverlesscarhq.com/solving-driverless-car-liability-2/”
    Here’s an article on some other potential solutions to liability.
    It should also be noted that in a bill just introduced into Colorado, the liability would be placed at the foot of the driver, which means the liability problem would be solved the same way it currently is – through insurance. And enough insurance companies have already indicated that they would be more than willing to insure driverless cars that this whole liability question is really just a waste of time.

  73. Matthew Newton January 23, 2013 at 2:15 am #

    Oh, and I got a source regarding Google accepting liability:
    http://www.cato.org/blog/googling-around-dc
    Turns out it wasn’t Brin OR Page, but Levandowski, who is the head of the self driving car program.

  74. Steven February 18, 2013 at 8:23 pm #

    Great.
    We are going to have robot cars idling or driving in circles on city streets because it’s cheaper than parking and the street is free. Cities are going to be absolutely clogged with robot cars and delivery pods going nowhere in particular.
    For the green urbanist future to come true we need car users to pay a market price for what they use. Of course conservatives and libertarians are only against government spending when its not their free stuff being threatened.

  75. Steven February 18, 2013 at 9:11 pm #

    Another thing, why is the future role of transit so black and white?
    I understand Brad’s point about transit energy efficiency, but that relates to fixed-route, scheduled transit.
    To solve the geometry of congestion problem what about dispatching mini robot buses when a bunch of would-be robotaxi passengers have similar destinations? The size of the group would conform to the optimal loading of the vehicle and this would in fact probably be more efficient than an individual car.
    I don’t see why smaller vehicles are always more efficient than larger shared ones assuming they are perfectly full. At best, imagine combining multiple small cars into one big one, and removing the sides and extra wheels.
    The choice to use this mode of transportation and accept the minor inconvenience of it would be for a lower price.
    Some people value speed or comfort more or less than others. Charging a market price for finite road capacity in the form of tolls would mean that in some cases a few users would pay a lot of money to reserve an entire block of roadway to travel at a higher speed and others would respond to the bid-up tolls by using less road capacity with shared vehicles.

  76. Doug K March 3, 2013 at 11:43 pm #

    I am surprised that it seems to be necessary for google to map where each lane is on each street. Can’t the car see the lane marking lines?
    As for jaywalking: Can the car differentiate between a legal unmarked crosswalk (at a local street intersection for instance), and a non-legal crossing? Can it understand that there are legal crosswalks at T-intersections? Does it know that in Portland it is legal to cross mid-block as long as you are 150′ from a legal crosswalk, but that the pedestrian does not have the right-of-way in that situation?
    I do look forward to merely stepping off the curb, in the dark, in the rain, with dark clothing on (like most pedestrians), and having the google car stop for me!

  77. John Niles (@JN_Seattle) August 8, 2013 at 8:10 am #

    There were presentations and a lot of discussion about vehicle automation and public transit at the TRB Road Vehicle Automation Workshop in July, pretty well documented at http://www.vehicleautomation.org.
    Has Mr. Walker written more on the topic?

  78. Kati Rubinyi December 12, 2014 at 3:16 pm #

    Great points you make, Jarrett. Our Autonomous Vehicle future will need to coexist with many modes and serve a variety of contexts for the benefit of our built environment and culture.