no, autonomous cars will not “abolish transit” in dense cities

Is transit headed for a collision with self-driving cars?  David Z. Morris in Fortune writes about how anti-transit Republicans are using the prospect of self-driving cars to argue against transit investments.

Alarmingly, he quotes nobody who can actually refute this argument, except in the fuzziest of terms.

Here is the recommended response:

We are currently in that phase of any new techno-thrill where promoters make grandiose claims about the obsolescence of everything that preceded them.  Remember how the internet was going to abolish the workplace?

In any case, technology never changes facts of geometry.  However successful driverless cars become, transit will remain crucial for dense cities because cities are defined by a shortage of space per person.  Mass transit, where densities are high enough to support it, is an immensely efficient use of space.

(Remember, a great deal of bus transit is in places where densites are not high enough to allow it to succeed; this is evidence of anti-ridership “equity” or “coverage” policies, not of transit’s failure.  Driverless taxis could certainly replace transit in those areas, assuming the pricing were gotten right, thus allowing transit agencies to focus on their core business.)

In many cases, people talking about driverless cars replacing transit are talking from an outer-suburban point of view, based on the experience of low-density, car-dependent places that are unsuited to high-ridership transit.  In those settings, if density is not increasing, they are probably right.  Driverless taxis will be more efficient than transit in these areas.

But all over the world, people are moving into dense cities, where even autonomous cars can’t replace a bus full of 60 people or a train full of hundreds.  There simply isn’t enough space to put walls between every pair of travellers, as the car model of transportation requires.  Nor will driverless taxis ever be there whenever you need them as great transit lines will.  Like bikeshare systems, they will experience surges where many of the vehicles are in the wrong place.

A city can of course choose to sprawl and avoid density to the point that driverless cars could dominate.  But in so doing it will fail to create a place that the 21st century economy will reward.  Real estate prices are already telling us that the market has chosen dense cities as the highest value form of development.  There is no dense city in the world that does not rely heavily on transit, for reasons of space-efficiency that none of the new technologies can change.  (Yes, autonomous vehicles will use space more efficiently than private cars do, but this is saying very little compared to what a great rapid transit network is achieving.)

Again, technology never changes geometric facts.  And the problem of cars in dense cities, and transit’s superiority there, is a geometric fact, a fact about space and its scarcity.

 

38 Responses to no, autonomous cars will not “abolish transit” in dense cities

  1. Ari November 4, 2014 at 10:02 am #

    There’s another issue with driverless vehicles in cities: when densities of pedestrians and bicyclists get too high, autonomous vehicles will have to pass a pretty impressive Turing Test to operate efficiently. Out in Mountain View (or wherever) algorithms for autonomous vehicles are (relatively) straight-forward: if a pedestrian is waiting to cross, stop for the pedestrian. If they’re at a light, stop if the light is red. This is all well and good, and probably works. These 99% of roads probably only encompass 1% of the potential driving situations.
    In dense cities, the calculus breaks down. There’s a left turn in Central Square in Cambridge, Mass which illustrates this point (and probably a similar intersection in any big city with a decent number of pedestrians). If the rules of a self-driving vehicle applied, it might never be able to make the turn. Nearly every turn at this intersection requires several pieces of information which are (at least now, and in the near future) much better processed by human contact. A turning vehicle’s driver often makes eye contact with several pedestrians with an ad hoc “you go now, then let me go” calculus. Often a driver will use pedestrians on an adjacent crosswalk to “run interference” and block that lane and edge in to the sidewalk. Drivers will make calculations: that older woman is less likely to start to run, while the kid with the basketball is more likely to do so. It’s a low speed dance that requires several bits of information to be processed but doesn’t use computer-type rules. This 1% of roads require 99% of the driving situations (but the real problem is that there are really infinite situations when non-computers—bicyclists and pedestrians—are involved). So while 99% of our roads might be compatible with self-driving cars, we might only be 1% of the way to having them.
    The issue for an autonomous vehicle is that its rules would, at rush hour, likely never let it make the turn. An autonomous vehicle would have a rule saying “don’t make the left unless there is no oncoming traffic in the adjacent lane.” Often the only way to make this specific turn is to wait until pedestrians in that crosswalk block the oncoming traffic, edge in to the oncoming lane, and then wait for a small gap in the pedestrian traffic there. An autonomous vehicle would have a rule that disallowed turning across a crosswalk unless there was at least a certain amount of space between the crosswalk and any pedestrians. In an intersection like this, that rule may not be valid for hours on end. An autonomous vehicle may assume a given speed and a given variation of speed for each pedestrian, while a human can make much easier judgments between children, adults and the elderly. An driver can see a slight head nod, or a quick wave of the hand, allow a gap to form for an auto to pass (this happens quite frequently). An autonomous vehicle will assume the pedestrian who has slowed their stride will just continue across at a slower speed and wait longer.
    When cars behave like cars on Interstate highways and suburban arterial roads, yes, autonomous vehicles are probably coming. But in cities, where cars often behave like humans (in all of the interactions above, the car is likely traveling at 5 mph or less) it will take a lot more processing power before they can function. At its core, an autonomous vehicle is still a 3000 pound hunk of metal, and they don’t interact that well on a human scale. If we all go everywhere in self-driving cars, then, yeah, they’ll be the wave of the future. But as long as people in cities walk and bike (which I assume will be the case) they’re not coming any time soon.
    (There are other physics involved with the 3000 pound hunks of metal argument, namely that it’s pretty hard to stop them quickly at high speeds when human beings may enter in to the equation.)

    • Alden April 20, 2016 at 9:44 am #

      Ari:

      I think your reasoning points in the opposite direction. It is in precisely the low-speed situations with large numbers of objects where botcars’ 360-degree sensors and high-speed processors will shine.

      Car/human interactions are unequal. As a pedestrian, I sure as heck can’t rely on eye contact to mean anything. The person behind the wheel of that 3000 pound hunk of metal may be an impatient jerk who thinks they can brush me back. So cars behaving like humans strikes me as a really bad idea. Cars behaving in a passive, non-confrontational, submissive manner to humans, possibly sitting and never making a left turn? Yeah, that’s how it ought to work.

      Your description of Cambridge Central Square sounds like a time and place where cars don’t actually belong. If botcars insist on dropping people off a couple blocks away, that is probably a good thing.

  2. Dan Sturges November 4, 2014 at 11:08 am #

    I am not sure why autonomous vehicles are being so rigidly defined as “cars”. There are many potential new forms of autonomous (or automated) vehicles or transit. There is also an opportunity to develop more localized systems, utilizing a smaller more optimized shape. And of course, there remains a large opportunity to develop dual-mode systems that travel both on the ground as well in a suspended minorail type fashion. There really is no benefit to constraining all our movement to only the ground plane. I have little doubt automated movement will ultimately replace public transit as we know it.

  3. Daniel Howard November 4, 2014 at 11:54 am #

    I like to think that in short order we will have smallish autonomous buses to serve suburban areas: we should be able to achieve higher frequency at a lower cost when the drivers are taken out of the picture.

  4. TachitoTec November 4, 2014 at 2:29 pm #

    I know a transport that you don’t need to drive, it was invented more than a century ago: train….car makers are bullshiting big time

  5. Rob November 4, 2014 at 5:27 pm #

    Jarrett, here’s another issue with driverless cars that amplifies your point: there will be a strong temptation for people with driverless cars *not* to pay for parking if they have the opportunity to have their car autonomously drive around the block until they’re done shopping, or drive home while they’re at work. When you get to less than single-occupant vehicles, the urban geometry issue is magnified.
    On the other hand, I do think there might be potential for driverless paratransit, combined with smart dynamic routing, to fill in for some niche transit functions (though not for the disabled, who need an attendant).
    A friend suggested decades ago that there are really only two key travel modes: pedestrian and driving. To me this has borne out well over time. Biking and transit work to the extent that they expand the range of pedestrians, and transit is only effectively in places where pedestrians thrive. So I’m curious how driverless cars would affect that proposition. Would they be tied to the car mode or straddle the two? I’m not sure.

  6. Ervin November 4, 2014 at 5:31 pm #

    Check out my new urban development blog…still a work in progress but give it a look…
    http://the703record.blogspot.com/

  7. Brendan November 5, 2014 at 1:15 am #

    There is also the fact that we are not actually close to developing driverless cars at all. We do not possess the mapping technology possible to allow driverless cars to work in the complex world outside of the short testing strips that they are currently tested on.
    You are living in a fantasy world if you think driverless cars are coming tomorrow. They are not. You will likely not see a driverless car in your lifetime.
    How can I put this any clearer?
    Our technology is not close to developing it. At all. The Republicans pushing the driverless car angle are rather wrong, disingenuous or both. Investing in the technology is fine, but don’t bank on it being the technology of your future. It wont be likely for a long time.

  8. Edward Re November 5, 2014 at 6:04 am #

    With questions of finite oil supply, pollution, obesity, global warming etc etc another question is why Republicans keep pushing cars and roads only? In a nation of 300 million possible car users, it’s going to be *very* expensive to keep following this path.

  9. simval November 5, 2014 at 8:45 am #

    The attraction of the driverless car is that of a magical technology that is hoped to be able to allow an unsustainable sprawling mode of life to postpone the day of reckoning.
    That being said, I will take the issue from another point of view and presume that the technology DOES become viable, which is a big jump of faith, whether it actually makes sense.
    The reality is that cars look economically efficient because of two factors:
    1- The driver’s labor is free as you are driving yourself
    2- They have no dead miles (if you don’t have to search for parking), every mile traveled is an useful mile
    Meanwhile, fixed route transit has one economic advantage, namely the ability to maximize vehicle utilization rates by picking up many people.
    Driverless taxis would destroy the two economic advantages cars have over transit while leaving the transit advantage alive. Though there wouldn’t be any driver, driverless taxis would still have to pay system administrators who provide the service and the cost of the sensitive sensor equipment driverless cars require to operate. So the driver is no longer as free as before.
    Furthermore, a mass adoption of driverless taxis would result in a lot of dead miles as vehicles drive empty from one user to the next, especially if used during peak periods where traffic is highly directional, or as they drive to holding areas and move around to distribute vehicles around the area to avoid situations where no taxi can be found in one area.
    This means that the costs of such taxis would likely be two to three times that of current cars. And indeed, if we look at cost profiles for taxis, we see that only around 50% of their costs go to the driver. So even removing the driver from the equation, driverless taxis would still likely cost 1 to 1,50$ per mile to recoup their costs.
    Meanwhile, if driverless technology can be put into cars, it can also be put into buses, where benefits will be much greater. Indeed, many suburban routes use needlessly big buses because most of the cost of running the bus is sitting behind the wheel, so small bus or big bus, it doesn’t matter. Driverless buses would probably cut costs to run buses by 30 to 40%, and allow for smaller buses (so-called midibuses) that are cheaper to buy and operate to replace large buses on many routes, also allowing for higher frequencies.
    The issue of space is also still true, with buses requiring much less space to travel around, a great boon in dense urban areas.

  10. Tom West November 5, 2014 at 9:47 am #

    Firstly, driverless cars can’t increae capacity in bumper-to-bumper traffic.
    Secondly, driverless buses could save 30-40% of costs – enough only to allow a nice boost in frequency.

  11. Patfromigh November 6, 2014 at 1:45 pm #

    The second chapter from Jarrett’s book:
    What Makes Transit Useful? Seven Demands and How Transit Serves Them
    1. It takes me where I want to go.
    2. It takes me when I want to go.
    3. It’s a good use of my time.
    4. It’s a good use of my money.
    5. It respects me.
    6. I can trust it.
    7. It gives me freedom to change my plans.
    IMO, autonomous cars will also have to answer these same demands to be successful. Sending autonomous cars to far flung suburban cul-de-sacs may not be profitable for those offering the service. The universe isn’t going to change for autonomous car services, the same geometry which dictates public transit efficiency will affect autonomous cars.

  12. Jase November 6, 2014 at 7:35 pm #

    You’ve made this argument so many times I think it might be time to spice it up a bit. Quantify it somehow. What exactly is it about cars that means we can’t fit them into cities?
    Saying “cars don’t work in dense cities” reminds me of a technology-based argument. And I know you hate those. Can we identify the features of cars that mean they won’t work? Is it having to fit space in between cars when they move? Is it the space provided per passengers? Is it the fact their paths must cross, therefore creating intersections and delays? Is it all of the above?
    To be honest, “walls in between passengers” is not a fact about geometry as about vehicle design. I can imagine efficient separation of passengers in a sci-fi scenario as easily as inefficient lumping in of passengers in an all-too-real circuitous bus scenario.
    This argument needs more robustness and more nuance if it is going to fly. And it seems it needs to fly to protect transit investment.

    • Mark November 1, 2016 at 4:03 am #

      I like your approach. There is plenty of room for many alternatives, who gets market share is dictated only by who most widely meets people’s diverse needs.

      That being said performance measured by space occupied is a sliding scale tied to performance based on speed

      400′ between cars is the safe following distance at 65 mph. Slower speeds take up less room. Furthermore 400′ is where an alert driver can stop safely100% driving more closely is common practice but involves risk

      400′ x 10′ wide road = 4000sq-ft seems costly for 100% safe travel at 65. So the dimensions of space, safety, and speed are inexorably linked.

      If door to door times for mass transit at 5 times higher for 100% safe travel then it stands to reason that the space argument is valid because mass transit carries more than 5 times more people

      However I believe this misses the point. People want speed and safety. Some more on one some more on the other. People other than city planners and transit advocates don’t care about land use. It’s mass transits failing to give people speed that results in low ridership.

      Land use should however be factored into cost to the driver or rider of all modes of transit. This means that higher performance comes at a higher cost. Borrowing 1000 sq-ft of pavement per second shouldn’t be free. Land in cities is expensive.

      Cars give people what they want. Until that is exceeded by mass transit cars will be here to stay.

  13. Zoltána Dosztál-Connell November 7, 2014 at 6:36 am #

    As a traffic engineer, I like to break down questions of technology like this:
    Take a critical intersection at the busiest time of day, and work out how many people move through it in one cycle of the lights. If you take any critical intersection in inner London, within very constrained geometry you’ll find several full buses carrying two or three hundred people between them on each 60-90 second cycle, plus maybe 30-50 people in cars and taxis, plus a large volume of pedestrians needing part of the cycle.
    The one important question about any surface transport technology is whether it can move those hundreds of people reasonably quickly without conflict, and how. Anything else is pure distraction.
    In the case of driverless vehicles: Yes, they can move that many people. How? By making those vehicles about the size of a bus, and making them behave like a bus.

  14. anonymouse November 7, 2014 at 10:16 am #

    If you make the vehicles even bigger, you get even better capacities. Or maybe, for added flexibility, you can even attach a few smaller vehicles together to form one big one so that they end up following each other at near-zero headway in a “train” of sorts.

  15. Jim November 7, 2014 at 1:16 pm #

    Perhaps the tone should be lighter regarding Republicans. There are some Republicans who support transit, and this kind of bashing is counterproductive. Mind you, many Democrats supported urban sprawl and freeway and oil subsidies including the current Great Oil Wars. If a transit advocate like Jarrett Walker comes across as anti-Republican, this only damages transit. And there are also many Democrats who support white collar transit (light rail, commuter buses, streetcars) for rich folk which deprives poor communities of transit. Both parties caused this problem called urban sprawl and the anti-transit climate of America.

  16. Jim November 7, 2014 at 1:20 pm #

    Also, let us not forget that soon our skies will be filled with drones delivering packages of all sizes (our meals, our groceries, our packages, shopping, etc.) which will greatly reduce vehicle trips of all sorts. But you do have to wonder how driverless vehicles transporting the young, elderly and disabled will impact transit. A senior community may hire their own fleet of driverless cars or vans to take their residents wherever they wish which would make an impact on both transit and para-transit.

  17. logan5 November 9, 2014 at 5:57 pm #

    How can anyone come to the conclusion that autonomous cars won’t replace buses or trains in dense cities without knowing the gains in capacity realized by autonomous vehicles? In my city, there is an even split in mode share between transit and cars entering and leaving the downtown core during rush hour. That means that autonomous vehicles would need to double capacity of city streets in order to replace all forms of public transit.
    Autonomous vehicles that are able to communicate with each other will be able negotiate signal controlled intersections far more efficiently. In the human scenario, the light turns green then the first car goes, then a couple of seconds later the second car goes, and so on. Autonomous vehicles will be able to move in a virtual train. The number of vehicles moving through an intersection should easily quadruple. Imo autonomous vehicles will “abolish transit”.

    • Mark November 1, 2016 at 4:32 am #

      It’s a fantasy that intersections will be obsoleted by communication between driverless vehicles. One error or accident can tip off a citywide pile up because all margin for error will be removed inside the intersection. And the instantaneous removal of an intersection will cause a good percentage of cars adjacent to the intersection to need to instantly reroute or face accident in adjacent intersections Remember in order to mesh vehicles can’t change instantly slow or speed up thru the intersection. All of those decisions need to be made by all the cars in the vicinity at once. This creates a reaction time that may be too long to avoid an impending crash.

      Terrorist will adore this if it comes to pass. It would merely take one manual driver or communications jammer to cause accidents throughout the city bringing it to its knees.

    • Norah December 18, 2016 at 3:21 pm #

      And what about people who can’t afford driverless cars? They may need to live farther from their jobs too, because their jobs are in the city and housing costs are often much higher in large cities. How will they be able to get to work if there’s no public transit any longer?

  18. Eric November 10, 2014 at 1:25 am #

    1) We should turn this logic on the Republicans’ head and say that since self-driving cars will increase road capacity there is no longer any reason to build new roads. (assuming there even was before)
    2) “Also, let us not forget that soon our skies will be filled with drones delivering packages of all sizes” – No, this will never happen because it is immensely wasteful of energy (not to mention dangerous). But there will be self-driving “shopping carts” which will perform the same function. You’ll select your groceries on a smartphone, they will be loaded into the “shopping cart” via a gigantic automated kind of vending machine which takes the place of a supermarket, and the “shopping cart” will drive itself to your house.

    • Mark November 1, 2016 at 4:47 am #

      1 there is 7 billion people now projected to be 10 billion in 2050. I think we will build new roads.
      2. Drone automated shopping carts sounds wonderful! At first it sounds like heaven. But then again maybe not. Wouldn’t you feel completely bored if you didn’t go out in public anymore except to work? As much of a hassle as shopping is it’s a nice break from sitting around on the computer or in front of the tv.

  19. simval November 10, 2014 at 1:13 pm #

    logan5, yes, and then we’ll have flying cars swooping around dropping people off.
    I am very skeptical of the idea of that scenario you put forth, it would depend on insane amounts of sensors and wireless communication and extremely tight programming. Any slight variation or communication loss could result in monster accidents as even the fastest computers won’t be able to deal with the deceleration needed to avoid accidents. Any such scheme can also not work unless every car is made autonomous, and would transform city streets into car sewers, which would be absolute hell for:
    1- City residents
    2- Pedestrians
    3- Cyclists
    There is also the issue of cost-benefit. The cost-benefit of autonomous cars is extremely low compared to every other alternative. The investments needed for the system to work would be extremely high. The sensors the Google car depends on cost 100 000$. Even if the cost was cut by 10, at 10 000$ per vehicle, the cost of sensors for the entire car fleet of the US would probably cost 25 trillion dollars, not counting the cost of retooling every single intersection into a smart supercomputer dealing with wireless communications with dozens of cars at once, nor the costs to upkeep the system. And all for what? To double the capacity (IF it works) when buses could do the exact same for a fraction of the cost?
    Oh, sure, autonomous cars would be faster… which would just incite even more sprawl and lead to even more wasted energy as people travel further and further.
    For 10% of that cost, you could likely give complete rapid transit systems to all cities above 100 000 people in the United States in the form of BRT, LRT or even subways.

  20. logan5 November 10, 2014 at 8:29 pm #

    @simval Volvo has tested car platooning on public roads. Each car 20 feet apart, at a speed of 52 miles an hour. This can be done because the vehicles can communicate with each other. This was on a public freeway, but how long do you think it will be before platooning can be done on city streets?
    Cost. Even at a sky high price estimate for each vehicle of 50 000 US, a transit authority could buy a fleet of 100 000 vehicles for 5 billion dollars. Assuming an average trip time of half an hour, and assuming the unlikely scenario that the vehicle travels or waits another half hour for the next customer, you get a capacity of 100 000 people per hour, plus door to door service. For that same 5 billion you get around 20 km’s of subway with less than half the capacity, and way worse service. Not only have you eliminated an expensive subway, you’ve also eliminated the buses to get you to the subway line.

    • Mark November 1, 2016 at 4:38 am #

      The problem with car platooning is accidents. If one car in the platoon gets into an accident all following cars plough into it. The removal of space between the cars is the safe stopping distance. Thus if there’s an accident with the lead car there is a guaranteed pile up of all 20 cars in the platoon.

  21. Eric November 11, 2014 at 3:10 am #

    “The sensors the Google car depends on cost 100 000$. Even if the cost was cut by 10, at 10 000$ per vehicle”
    The cost can be cut by much more than that if the vehicles are mass-produced. Google car sensors consist of sonar devices, stereo cameras, lasers, and radar. None of these is intrinsically expensive, but in the experimental phase they have to be custom made. A cell phone you’d pay $100 for now would have cost millions of dollars a couple decades ago, and we will see similar progress once car sensors are mass-produced.

  22. Jack November 11, 2014 at 7:46 am #

    @logan5 But now you’re right back to the geometric problem Jarrett highlighted.
    Let’s make your example a little more concrete. You’re suggesting that if DC had the option of expanding the Metro system with a new 12 mile line across downtown, the area would be better served by buying 100,000 self-driving cars. Ok – as has been asked repeatedly, where are you going to put them? To park them all day would require bulldozing a square area from the Mall to K Street and from the White House to the Verizon Center. You can do that, but then you look like Detroit with massive gravel lots creating vast parking prairies on the edge of downtown. Even doing it all as five story garages ends up costing you about 15(!) city blocks. Somehow that seems less desirable than the type of city Washington has built around the Metro these past 40 years, not to mention the monetary cost of such massive construction (and destruction of property values).
    Oh, they’ll just autonomously circle until someone needs them? Well, those 100,000 commuters won’t need them again until rush hour and there’s very low demand in the middle of the day, so now divide the above space out over the city’s roads to figure out how much room they’ll be taking up aimlessly tooling around wasting gas, polluting, getting in non-motorists’ way, and congesting the streets for other drivers.
    You’re suggesting that either of the above should be done so that ~1/8 the current ridership of Metro could take a car door-to-door instead of walking from a transit stop. I’ll take the new subway line, please.

  23. simval November 11, 2014 at 10:43 am #

    logan5
    And I saw prototypes of flying cars too. There is a huge difference between a test made under perfectly controlled conditions and reality (even then, I remember a German company showing off an obstacle detection system in their lab, in front of the media, where the car just plowed right into the obstacle without even slowing down, oops, fail). Platooning on grade-separated freeways may sound nice (then again, pray to your god that no tire ever bursts, nor any error ever happens, because if just one car swerves off, you get a 20-, 30-car crash no matter how fast the computers are), but the idea of putting that on city streets is absolutely terrible, especially as cars would continually want to integrate and leave the platoons, forcing other cars to move away. And forget doing anything like that as long as you have pedestrians, cyclists and even driven vehicles sharing the streets.
    Unless you make all city streets into freeways, your vision is impossible. And what a nightmarish, dystopian world that would be one in which every city street is turned into freeways.
    If driverless vehicles ever become a reality, driverless buses would be much more efficient than driverless cars. No door-to-door travel and that’s a good thing, fixed routes lead to better vehicle utilization. People walking around contribute to public health, to quality of life and to fostering life and commercial activities on streets. Door-to-door motorized travel isn’t an ideal, it’s something to avoid, it completely destroy the public realm.
    @Eric
    Processing power does increase and become cheaper, but precision and durability remain expensive. The problem isn’t the computer, computers have had the processing power to drive vehicles probably since the late 90s. The issue is that computers are garbage-in-garbage-out, their outputs are only as good as their inputs. The sensors are the weakest link, they not only need to be able to detect things around them on a clear summer day, but during all weather conditions, and do so reliably for years with great accuracy, even if they’re exposed to elements all year long: from snow and glacial cold to scorching sun and great humidity. When you’re driving a couple of tons of steel at 60 mph, 90% accuracy is not enough, 95% accuracy is not enough, 99% accuracy is not enough!
    Even Google’s car, worked on incessantly by an army of specialists, with all the expensive equipment they can find, cannot tell the difference between a paper bag and a big rock in the middle of the road. Imagine a platoon of cars driving along at 70 mph on a freeway, a discarded newspaper flies across them and the cars swerve left and right to avoid it as if it was a huge rock or a tree in their way, deciding that colliding with each other is better than hitting the obstacle head-on. Some cars brake better than others and suddenly they’re rear-ending each other.
    You can’t just strap a couple of 40$ webcams on a car and call it a day. Accuracy, reliability and durability are expensive, extremely expensive.
    For instance, traffic lights programming is extremely simple any smartphone could run it, but traffic light controllers still cost around 10 000$ to buy. Because that’s the cost of knowing the computer will not fail even exposed to elements for years on end.
    If the technology happens, it happens. But sacrificing the present hoping for a magic solution to come down the line and solve all your issues without you having to do anything or change your habits is not a wise course of action at all.

    • Mark November 1, 2016 at 4:51 am #

      Well said

  24. Sandro Wolfe November 28, 2014 at 8:33 am #

    how about this concept od shared city spaces ?
    http://smart-magazine.com/space/the-miracle-of-space/
    maybe we finally have less cars, more pedestrians, more cyclists AND less pollution AND a better community spirit ???

  25. Kisai December 22, 2015 at 12:49 pm #

    In my opinion, driverless cars will replace taxi’s/uber not personal cars. Here’s why:
    A) the technology solution, requires all the driverless cars owned by one company (or all companies) to communicate with each other, creating their own version of moving block communication controlled vehicles, and like driverless trains, the cars will not operate unless it is safe.
    B) the taxi company will be responsible for running their own Taxi OMC to ensure that the vehicles are safe
    C) the vehicles themselves would likely have to be electric-only and return to the OMC for charging without some kind of automated queuing charging network.

    Personal cars require retaining parking spaces, which are not a good use of space.

  26. Thomas Robertson January 5, 2016 at 1:38 pm #

    “Cars” as we’ve understood them are probably going to go right out the window with the self driving vehicle revolution. What’s more likely is 4-50 person autonoumous vehicles which opperate as dynamic paratransit. I request a ride to work and I’m told to walk a few hundred meters to a major intersection where I meet 10 or so other commuters around my neighboorhood. We ride over city streets to an area close to where we’re all going and walk the rest of the way. In suburbs the vehicles are smaller, not too dissimilar to old school “cars”, but in dense areas the trend is towards 10-20 person vehicles (smaller versions of what currently are city buses). That seems quite efficient even for very dense places. My experience is living in Boston and if I wish to walk to the train and from the train to my office, it’s an hour commute. Meanwhile a solo driven car can do it in 40 mintues. I would wager that if we forced the solo driven cars off the roads and replaced them with 5+ people to an autonomous vehicle, we could replace the public transit system with one that doesn’t require public money (though we probably should give subsideies to people to help them afford the cost of a ride if they need it).

    While on-route to that future I mention above I’ll make another prediction. We might at some point pave the tracks of our existing subways and run back to back 20 person automated bus’s down the track. That could exist long before autonomous cars, and the vehicles could be standardized for sale to all kinds of metro systems, rather than being very specific to the transit system the way existing vehicles are. Lord knows the Boston T could use more reliable and chepar vehicles with all the winter maintenance woes they’ve had.

    • Mark November 1, 2016 at 5:03 am #

      Shared car cabs worked well in Russia during the soviet collapse. Very low price to get around and street to door transit. Rumored to be the seed idea behind uber. Now with uber and lyft door to door is possible. With driverless vehicles only thing that changes is price. What’s a car & driver worth vs what’s the driverless car worth?

      • Norah December 18, 2016 at 4:00 pm #

        Not everyone can afford Uber-type pricing for their daily commute.

  27. Alden April 20, 2016 at 9:46 am #

    ‘Autonomous cars’ is a mouthful. I predict it will go the way of ‘velocipede’, ‘horseless carriage’, and ‘automobile’.
    Let’s call them ‘botcars’.

  28. Roderick Llewellyn June 11, 2016 at 1:51 pm #

    Proponents of robocars point out that the current car fleet spends most of its time parked, leading to wasteful land use… which is true. They then say that moving to a robocab model in which you call a car when you need one will be more efficient. However, the same land use that makes transit efficient – namely high density – is also what you need to make wait times low and efficiency high for shared vehicles. Also, during rush hour, the size of the total fleet can’t really be significantly smaller than the current fleet, since most people are traveling about the same time. But having a large percentage of the population using robocars rather than transit doesn’t actually support a dense land use which needs walkable neighborhoods, not derivable ones, even if a robot is behind the wheel. Thus, to make robocabs work really well, you need a high percentage of people using transit.

    • Roderick Llewellyn June 11, 2016 at 1:52 pm #

      That should have been ” not driveable ones” 🙂