Moving all these light rail riders into autonomous cars will help solve traffic congestion, per automobile technology investor. (Photo by Oran).
Just when you thought silly season was over for transit opponents (We should vote down ST3 because Sound Transit threw a large opening day party for U-Link!), Bryan Mistele, CEO of INRIX, a traffic-information and connected-car company based in Kirkland, has penned a ludicrous non-sequitur in the Seattle Times, arguing that the driverless car of the future will make transit obsolete ($), and, oh yeah, eliminate traffic congestion.
It was timely that the Times ran Mistele’s piece so soon after the first road fatality involving a car running on auto-pilot was announced. Live Science recently poked holes in the claims of the relative safety of auto-pilot technology:
In fact, a study published in October 2015 found that self-driving cars are more likely to be in an accident. The study, conducted by the University of Michigan’s Transportation Research Institute, found that per million miles traveled, self-driving cars had a higher crash rate than traditional cars. At the time of the study, no self-driving cars had been found at fault for the crashes they were involved in.
The technology will improve. Cars will learn to optimize their following distance (but that means that they will be 4 seconds apart, which drivers today are too impatient to do, so that means less road space available). Car computers will learn to talk to each other (hopefully better than Microsoft and Apple), and software bugs will be worked out so that computers don’t “crash” while they are in the driver seat. Everyone will want one, and be able to afford them. If you believe all that, I’ve got a bridge to sell you.
But even if all those improvements are possible, the broader claims about congestion reduction and transit replacement are real whoppers, a one-two punch of Deus ex Machina that fails to show its math. CleanTechnica offers a more sanguine, and partially data-driven, analysis.
Congestion is a serious issue in developed countries, but autonomous cars are likely to be as much a part of the problem as part of the solution.
The article covers the likelihood of induced demand from autonomous vehicles, consumer attachment to private vehicles they can get in anytime, the non-linear relationship between vehicle miles and congestion, and the various cases of how autonomous cars will merely add to vehicle miles. And of course, it’s intuitively clear that removing pain points to driving – congestion’s not so bad if you can sleep through it – will only further incentivize passively occupying travel lanes.
The claim that autonomous vehicles will render fixed-route transit obsolete is particularly unfounded, with basic geometric facts providing the reality check. Yet nonetheless the argument has become a trendy political talking point, as Fortune documented back in 2014. Generously assume that “small form factor” vehicles succeed in doubling vehicle throughput capacity (a big if!). Then assume a standard vehicle occupancy rate of 1.5. Assuming these two factors, the capacity test for autonomous vehicles as congestion reducers and transit replacers is whether or not transit could reliably carry more than 3 people in the same space. That’s a laughably low bar for any urban transit agency. And for a central city like Seattle’s, with 35% of people already taking transit while using 10% of the space? Any major transfer of people from transit to small autonomous vehicles would represent a loss of capacity, not a gain.
A further problem with autonomous car technology is that it is being designed around the paradigm of maximizing efficient and safe flow of cars around each other (and talking to each other, presumably using a master protocol agreed to among all manufacturers). There have even been calls for removing traffic lights from city intersections to provide continuous flow of autonomous vehicles, to the direct detriment of people walking and biking. A team at MIT went so far as to model how autonomous cars could go around each other at intersections, and forgot to add pedestrians and bicyclists entirely, as chronicled by Citylab.
From the get-go, designers of autonomous vehicles have focused on replacing human-driven cars. Those who extend that mission to replacing bikes, buses, trains, and walking have missed an important point: Driverless cars can take up less space than human-driven cars, but they just don’t take up less space than crowds of pedestrians, bikes, buses, and certainly not grade-separated trains stuffed with hundreds of passengers. Smarter investors will sense an opportunity to apply the technology broadly to transit, and work with transit agencies to improve automatic train and bus control technology rather than antagonize these agencies based on poorly-thought-out ideological notions of forcing everyone into cars.
This is not to say that light rail will help eliminate traffic congestion. It will certainly help speed up commutes for those who ride it, but freeways are going to continue to get more clogged as population grows, people move further and further away from their jobs, and demand is induced any time road capacity is freed up, at least in an urban environment. Light rail will at least mitigate the rate at which congestion gets worse for those who continue to commute on the freeways, by moving tens of thousands of commuters off of the freeways.
Nor is this to say driverless cars won’t play a positive role in the transportation system of the future. But to say they are a reason to vote against high-capacity transit systems (or buses, or investments in bike and pedestrian infrastructure) defies simple spatial math. And besides, fully grade separated transit provides massive new capacity while leaving the street grid fully available for technological experimentation. Autonomous car investors should be grade-separated light rail’s biggest fans.
In many ways, I think driverless cars will actually end up increasing the demand for transit by providing the last-mile service at a relatively cheap cost. They can also serve as a viable replacement for transit in areas with exceptionally low ridership (e.g. Sammamish, North Bend), allowing limited tax dollars to be re-directed into corridors with more ridership potential.
In the meantime, increasing traffic congestion will increase the public pressure to allocate dedicated space for transit, and routes like I-405 BRT will suddenly become a lot more usable if driverless cars can be cheaply summoned to connect to it, and in the city center, people will continue to ride transit for the simply because it will be faster. On the contrary, giant parking garages at transit centers will become while elephants of an earlier era, as the cars simply drop off their occupants and head on there way, rather than sit at the transit center all day.
To add to that if people stop owning cars and start calling “shared/rented” driverless cars in an Uber/Car2Go/ReachNow kind of world they will start to fully understand the cost of driving, and more then that some of the costs of externalities such as congestion, pollution and crashes (which should be greatly reduced) can start to be more accurately built into the cost of use whether that be per mile or per minute.
This will give consumers a better sense of what it actually costs to drive vs. bike, walk, or take transit and may push more people to alternative modes.
Plus, people who switch to use shared driver-less cars will also be more likely to use traditional transit, as mass transit will be be preferable in some situations. For example, during peak demand there may be Uber-like surge pricing, causing more people to switch to mass transit.
Agree with both comments. As the TheUrbanist article detailed, routes like 405 BRT can actually become MORE productive than light rail but to the ability to flex vehicle sizes throughout the day and have super high frequency all day when labor costs are removed.
But for the purposes of ST3, building light rail remains the better choice because the fully grade separated ROW remain an effective hedge against “BRT creep.” Perhaps come ST4 we will be primarily investing in BRT corridors to extend the network because of the ability of autonomous vehicles to boost the productivity of those corridors … but we are not there yet, not close.
Converting lanes to “autonomous-only” will be as politically difficult as giving lanes to transit only, and driving self-driving cars in mixed, urban traffic is going to have the exact same congestion issues as current technology … any gains in efficiency will be offset by induced demand by people more tolerant of congestion due to the fact they aren’t driving anymore.
“Converting lanes to “autonomous-only” will be as politically difficult as giving lanes to transit only”
It’s more like electric cars. Electric cars get some tax breaks, and have no equivalent to the gas tax (although they pay a fixed fee to offset some of that), and I think they can drive free in the HOT lanes if not here then somewhere. These incentives may be worthwhile to jump-start a more environmentally-friendly car base, but they can also end up being subsidies for the rich. Because it won’t be the working-class person who can barely afford their MVET who will be buying new driverless cars, and used ones won’t exist for a few years, and cheap used ones won’t appear until a decade after that (if they ever do).
“routes like 405 BRT can actually become MORE productive than light rail”
If productive means high capacity, rail has intrinsically higher capacity because you can couple vehicles together for zero following distance and less energy required. That doesn’t necessarily mean our existing trains have the maximum theoretical capacity (especially our streetcars) but a change of vehicle could easily provide it. Link’s cars have a driver compartment at both ends, which is useless when an inner end is coupled, but some of Portland’s MAX cars have seating areas instead of driver’s compartments for the inner ends. (And of course driverless trains wouldn’t have a driver’s compartment. Although it may need a reserved staff seat of some sort. What does Skytrain have?)
I’m having a hard time thinking of other ways a bus can be more productive than a train, unless you’re thinking of open BRT, flexible routing, or lower cost. But I wouldn’t call those “productivity” per se. Open BRT may be more useful in another way, by serving a larger area, but it would only work if demand in the shared segment doesn’t exceed the capacity of those open BRT lines going to leaf areas.
>> But for the purposes of ST3, building light rail remains the better choice because the fully grade separated ROW remain an effective hedge against “BRT creep.”
How so? The only reason our light rail is going to be grade separated is because the folks building (ST) says it will. They have reneged on similar promises in the past (First Hill station). With that being said I agree that grade separation (or at worse, a little bit of surface running through Interbay, which is quite reasonable and fast) is highly likely.
But there is no reason to assume that a similar investment — if specifically called out — in bus infrastructure wouldn’t achieve the same results. We built a bus tunnel many years ago, and not one general purpose car has driven on it. So far as I know, the same is true for bus ramps and bus lanes throughout the region. Even HOV lanes, as flawed as they are, haven’t degraded from a regulatory standpoint. I think that is normal. I think it very rare for an agency to build a bus lane, and then decide to turn it over to cars.
“BRT creep” refers to a practice that is common amongst all modes. An agency promises something great, then under delivers. It usually is about money. Sound Transit did exactly that. Not only with the light rail line, but with the streetcar (both rail).
If ST promised to build a bus tunnel, or an Eastgate to downtown Bellevue busway, my guess is they would build it just as well as they would a train tunnel (or railway).
The difference is that now ST seems to recognize that grade separation is important and there’s a lot of public demand for it. That only happened because many people kept saying repeatedly for years that the cost of grade separation is worth it, and it spread from just a few transit fans to a wide cross-section of the public. ST2 was fully grade-separated at one point and it was slightly reverted only because Bellevue wanted its downtown tunnel and asked ST to economize elsewhere. Lynnwood Link in Shoreline has underpasses, and south King County will have that too, and beyond those it will be elevated. The only surface segment contemplated for Ballard is on 15th Ave W where the road has exits rather than intersections.
I agree with asdf2. Driverless shuttle vehicles are already being considered for last mile connectivity in several countries.
It may shift transit route design significantly, too. Agencies will consider how to create fewer, higher-speed, higher frequency routes and let autonomous shuttles do the less productive and less cost-effective routes in the short term — or a person’s own vehicle will be sent home after dropping them out nearby at a station in the long term.
In any case, technology is increasing the role of drop-off and pick-up at stations even now (text messaging, Uber and Lyft) – and this trend is ignored by many station designers and local city traffic engineers.
It’s really unfortunate that Local Motors is aiming its 12 passenger self-driving bus as a transit alternative rather than a transit compliment.
https://localmotors.com/olli/
It could be an ideal feeder vehicle.
Easymile markets itself for rail hubs.
http://easymile.com/mobility-solution/
“The technology will improve. Cars will learn to optimize their following distance (but that means that they will be 4 seconds apart, which drivers today are too impatient to do, so that means less road space available).”
I’m not a fan of autonomous vehicles being able to save us but I cringe at this. It is more like a 2 second or a 3 second rule depending on which state you are in. A good portion of that time is because it takes time for humans to process thought and get their foot applied to the brake. Computers *should* be able to do this faster than humans making this a shorter rule of thumb for autonomous vehicles.
It’s widely assumed in the highway engineering world it takes 2 seconds for a human to process/react/notice whats going on and another 2 seconds for the vehicle’s mechanical systems to brake hard w/o locking up and come to a complete stop. Realistically, any engineering signing off on this wouldn’t reduce that safety buffer below the 2 seconds in the event of an unforseen issue (mechanical failure, medical emergency, collision, etc).
In our rainy conditions, performing a full 60 mph stop will take almost 5 seconds fully automated. Automakers are definitely going to be coding things up on the ‘cover-their-asses’ end of things. You have to first run for congress and actually accomplish tort reform.
The whole focus on coming to a complete halt with regards to following distance is ridiculous. It is impossible for the car ahead of you to come to a complete halt instantaneously, unless there is an immense obstruction in the road, which is extremely unlikely.
Its roughly 3 seconds fully automated for 60 to 0.
http://www.edmunds.com/driving-tips/keep-your-braking-distance-more-than-just-slowing-down.html
Wouldn’t an autonomous vehicle be on the road during their “empty” trips too? That would seemingly increase VMT and thus negate the effect of better lane utilization.
I think you’ve hit the nail on the head – there’s going to be much more congestion with self-driving cars. Rather than parking right at the destination, why wouldn’t a driver / car owner / user [what’s the right terminology for one who owns and uses an autonomous vehicle?] just send the car to find a less expensive place to park (maybe even return to home). For short stops, the owner might just send the car to “orbit” until summoned. Either way, it’s more cars.
It seems the assumption is that self-driving cars will be shared use, a la Uber. I would think that anyone who’s been in a parking garage or transit station elevator will be quickly disabused of that notion the first time a litter-filled car reeking of urine pulled up at the curb.
“just send the car to find a less expensive place to park (maybe even return to home)”
Where is a guaranteed free place to park? At home, or in the suburbs if they live there. But even in the suburbs it will be hard to park more than a few cars in a store’s parking lot or in a public recreation area or on the street without people complaining, so people will find there’s not a lot of places to store the car except at home, unless free public lots for idle autonomous cars become a thing.
On any given workday, Seattle has tens of thousands of unoccupied, unmetered on-street parking spaces within a few miles of most employment. Some are in neighborhoods, some are in lower-density commercial areas. But unless the law is changed, any of them would be suitable parking for a commuter car.
Even now, with human drivers, if you look you’ll see lower-budget commuters parking just outside of time-limited parking and walking the extra distance into downtown or to a bus stop.
If you’re just driving downtown from one of the hills, sure, your car might well go back home. If you’re driving in from Renton or Issaquah, though, there are plenty of reliable, closer-in alternatives.
Eventually someone will build an app that monitors parking, and will send your driverless car to the nearest relatively empty shopping center lot.
Ed P
It is very simple to have an Uber-like rating system, because you need to tap or use an app to pay anyway. That will stop people from wrecking the cars.
A Ford Excursion following a Mazda Miata might want more than just 2 seconds following distance :-)
Or a pickup loaded with unsecured water heaters
Every time I see the “driverless cars will make transit obsolete” article, it reminds me of the people in the 90s who said that we didn’t need to invest anything in road or transit infrastructure because, hey, in ten years, everyone would be telecommuting.
You mean the very same sort of futurism that delayed our region’s transit investments for 30+ years until we came to our senses?
Or decades ago when it was proposed that many would be in home helicopters and not cars!
The most depressing thing is that some people are seriously pursuing this as a reason to oppose transit investments.
When you’re faced with an uncertain future, the prudent thing to do is to hedge your bets by investing in a reliable fallback, which is what transit is in this case. If you don’t, you’re trusting in the unknown and seller’s promises and may end up with nothing.
In the transit community, conversion to automation is still a struggle for vehicles that run on rails in very controlled environments (see NY Subway). Even the “relatively simple” task of implementing positive train control (PTC) on our nations freight and passenger trains has come at significant cost and delays due to tons of unforeseen and technological challenges. If installing semi-automation on 10,000 locomotives and 150,000 miles of rail has been a decade-struggle for six companies; imagine how difficult it’ll be to outfit/covert 200+ million vehicles with another million miles of road used by 300+ million people.
And don’t forget the people who refuse to give up the wheel. How do we deal with those when the law of “it’s illegal to touch the controls of a car” inevitably has to be made for this system to fully work as intended?
And don’t forget labor relations. Transit driver unions will object to automation that costs jobs, which will delay the implementation of driver-less transit.
[I’m not saying this is good or bad, just agreeing with Mike that the time to implement these technological solutions is wildly optimistic]
I suspect PTC delays are less about logistics and more about railroads not wanting to pay for the technology, which comes out of profits.
“Even the “relatively simple” task of implementing positive train control (PTC) on our nations freight and passenger trains has come at significant cost and delays due to tons of unforeseen and technological challenges.”
The delay isn’t due to “unforeseen and technological challenges” – it’s money and politics. The private railroads lobbied against it because they don’t like being told how to spend their money, and have managed to push back the deadlines for getting the job done; on Amtrak it’s been delayed because conservatives in Congress don’t want to allocate the funds.
I’ve read the suggestion that autonomous cars will make parking obsolete since they can just circle the block for “free”. Imagine what that will do for congestion.
I think most city drivers find it necessary to bend the rules from time to time in order to make the system work. What will be interesting will be to see how autonomous cars are programmed. Do they take a firm view of traffic laws or do they program in the same pragmatism which most human drivers employ? Or, the sheer recklessness of scofflaws…
Plus the resource consumption require to keep vehicles constantly moving; occupied or not.
Autonomous cars won’t make parking obsolete, but they’ll make current parking garages obsolete.
Autonomous cars don’t need walking-height ceilings (though autonomous SUVs would still need that height), or room to open car doors once parked. They don’t need wide aisles for sloppy human drivers to get in and out of spaces.
When autonomous cars reach sufficient mode share, garages specifically catering to them will be able to pack in significantly more vehicle density, lowering cost per vehicle, leading to a premium on the price of parking for human-driven vehicles. This will create one more incentive to retire human-driven cars, at least in cities where parking requires expensive structures.
So, you’re talking about parking garages that only admit unoccupied cars, then?
Exactly. Why would you want to go to the garage yourself if you could go to the front door of your ultimate destination. Town cars don’t make you get out at the garage entrance, the driver takes you to the front door.
Human parking will be a premium service at a higher price. Your car will let you off right at your intended destination, then find itself a parking space.
“Do they take a firm view of traffic laws or do they program in the same pragmatism which most human drivers employ?”
If a human driver does it, the government is faced with millions of individuals it can ticket only if a police officer at a particular moment sees probable cause for an unsafe situation. If an autonomous car does it, the government can easily sue the company for intentionally programming that into its algorithm. So no more speeding, circling the block several times, etc. They may be able to get around it by circling a larger area.
A really smart car could just give itself a ticket, and debit some Oops account on-board.
We encounter many untruths in our lives. Sometimes they are things people want to believe because it makes them feel better about who they are. Sometimes they are lies propagated by those who stand to gain. Most lies do not stand basic thought or light research. If people just asked, ok, wait a sec. How does this work? If they just asked, is this true, they wouldn’t be fooled.
Self driving car myths probably has components of both. Every time I see an article talking about self driving cars helping congestion, it is written by someone invested in the business. People don’t like being shamed for causing congestion, but they like driving in their car. The idea that there car could get better so they could keep using it without causing congestion is something happy. The problem is of course, anyone who thinks about it for a bit will realize it is complete bullshit.
I wrote a comment a while back after seeing a few of these self serving car claims. I just go through and say lets think about what is actually likely to happen. I might be off on some points, but I know there is no reason to believe people with the biggest motivation to lie are right.
The idea of congestionless driving goes back to utopian film clips and propaganda, like as the Futurama exhibit in the 1939 world’s fair, and TV shows that modeled suburban life without addressing the inevitable congestion that would arise. (In those TV shows, 95% of the action is in the house, and most of the rest is at work or school. They show the car driving out of the driveway, but not on the road where real drivers would encounter congestion. And they don’t show the weekly errands that real families have. How does the Brady Bunch refrigerator get refilled?)
I’ve also seen an argument that when Los Angeles went highway-happy in the 1920s, there was a generation-long period where the boulevards and freeways really were wide open until growth caught up with them. So Angelinos have a cultural memory of wide-open roads and expected that they’d remain that way forever… because isn’t that what the propaganda films promised? Meanwhile New Yorkers, according to this report, “never entertained the notion that they could drive from 110th Street to Fulton Street at 50 miles an hour” so they don’t have that memory or expectation.
Seattle also had only 200,000 population when the Aurora Bridge was built, and only 10% of people had cars although that number was rapidly increasing. Even when the Forward Thrust subway failed in 1972, Pugetopolis had less than a million people and most of them lived between Bothell, Redmond, and Kent, and a lesser contingent to Lynnwood. So in that era I-5, 520, I-90, and 405 looked like all they would ever need. (The original 520 tolls went off seven years early because the Eastside grew much more than expected as 520 was extended.)
While preaching the benefits of a “smart intersection”, in a pretend world where bikes and pedestrians do no exist is bad, the concept of a “smart intersection” need not be inherently detrimental to the pedestrian.
For instance, one could imagine a design with two signal phases – one where the “walk” sign is on in all directions, and all the cars wait, and another phase where the pedestrians wait and the cars do their autonomous negotiation, allowing a single signal phase to take care of movement and turning in all directions. (Bikes would presumably cross during the pedestrian phase). If done correctly, I don’t think this would be any worse for pedestrians than the traffic lights we have today. Of course, the best configuration for pedestrians is to not have traffic lights at all, but narrow streets intersected with four-way stop signs.
I like it asdf. And also, this would allow the signal times to be much shorter, meaning not having to way so damn long on the sidewalk.
To be clear, I’m not sure it would be any better for pedestrians than now, just that I can see it not worse. Picture a junction like Market/15th today, with lots of long left-turn phases, and any given direction has a walk signal (or flashing “don’t walk”) maybe 10% of the time at best. For intersections like this, it’s hard to imagine a two-phase “robocar/ped” signal being any worse to cross.
Pedestrian underpasses would allow intersections to have no lights. That has been done on a modest scale in Russia and England, but the English ones aren’t very popular. (In England they often have both an underpass and a crosswalk, so most people use the crosswalk to avoid walking down and up, and the underpasses get an unsafe reputation. In Russia everybody uses the underpasses because it’s the only way, and they have ramps rather than stairs, and some of them have a metro entrance and small shopping center at the bottom. I thought they were a good idea.)
The problem with a pedestrian scramble (all-way walk) now is that it can give pedestrians only 1/4 or 1/6 of the time (especially if left turns each get their own half cycle). Some have argued that scrambles are pedestrian-unfriendly because with a regular intersection there’s always one direction open whereas with a scramble you have to stop completely even if you’re going kitty-corner and can go either direction. A lightless scramble with pedestrians getting half-time might be better, but I doubt it would really be half. It still takes time for a lot of cars to go every direction even if they can interleave.
Those are placed in much denser environments than we have anywhere in this region. They are expensive to provide, take up land and how many of them would you think you’d require to accommodate self-driving cars in downtown Seattle? Two hundred? Give me your capital budget and explain how you’re going to finance them. Oh, yes, and security. And drainage (it will continue to rain here after self-driving cars appear).
I’ve seen Google Street View pictures of the Russian underpasses and they look horrible. Going down the stairs, passing through a narrow, very crowded underground walkway, then going back up would likely take longer than waiting for all but the worst lights in Seattle – and all in the name of free-flowing motor vehicle traffic.
They’re only at major intersections, not every single block.
Theoretically, couldn’t a pedestrian or cyclist just go out into an intersection whenever they want? Presumably autonomous cars would be programmed to avoid running into them…
That video of the self-driving car intersection without traffic signals reminds me of this Onion video:
http://www.theonion.com/video/obama-replaces-costly-high-speed-rail-plan-with-hi-18473
The bus shoulder lanes are actually happening on I-405. We desperately need them on I-5, especially to get through the reverse-peak-direction parking lot.
As for the high-speed buses, if autonomous technology can work for cars, certainly it ought to be able to work for buses. Mr. Mistele, can you make this Onion video a reality?
Brent;
Great piece. I totally agree with your thoughts.
I’d like to add one of my own – namely that automated vans like Olli ( http://meetolli.auto/#firstPage ) can help with paratransit. Imagine if at least some paratransit clients could hail an Olli and use that to catch up with a fixed route or light rail or even get to a Point B without the large costs of a transit operator? Would really help transit agencies out as paratransit co$ts are really climbing…
Thanks;
Joe
I can see Olli filling lots of use cases, but not most paratransit use cases. There is securement technology to consider, the need to provide quick medical attention (higher among paratransit riders than the general population), and plenty of cases of vulnerable riders who still need to have a trained, accountable companion with them.
I honestly see more opportunity to get some paratransit riders using the train for long-distance trips, possibly with a travel assistant in tow, and maybe with a staffed Olli solving the last-mile problem on each end.
Thanks Brent, I’m just trying to think of win-win solutions :-).
If the autocar frees up the all-day assistant from driving, that could be more pleasant for them and allow them to be more assisting.
Olli is staffed. If you look through the descriptions of it it is move of a remotely monitored and controlled vehicle than autonomous. They don’t say how many vehicles per staff member they expect to achieve.
Good comments guys, thanks.
Hundreds of thousands of individual machines, under very hard wear, operating in close formation, at very high speeds. Manufactured by whom? What’s safe inspection schedule- once an hour? Same for checking, let alone changing, tires. Done where and verified how?
And behind it all, whatever training, skill, and reflexes the dumbest human needs to take a car down the driveway- an automated system has to be told literally everything, to many decimal places’ precision. Meaning how to see, hear, feel smell and react correctly to events and conditions changing by the second. As dictated by the most limited human delivering the last keystroke or mouse-click.
Couple of really simple principles here. One, by professional ethics, an engineer will tell you what choices are technically possible, the mechanical advantages and problems, and costs, of each. Except for piece of information: “What SHOULD I do?” “Artificial intelligence” is a mass casualty squared circle.
The world’s worst human driver has at least a billion years of survival experience designed into their every cell. Including senses and reflexes, maybe most of them, we don’t even know are there, or how many times every day we would’ve been killed without them. Which add up to our real survival advantage: we care if we get killed.
Main thing I’m hearing here is excuses from people too lazy to learn to drive, or scared to try Which several decades of reliance on part time drivers prove can be inculcated along with survival reflex of not reporting 59 seconds late. Tempted to say clone my trolley instructor. But..There Are Some Things Mankind Was Never Meant to Know!
Mark Dublin
Google’s engineers think we won’t have widespread automated cars until the 2050s for these reasons.
So what will we do until the 2050s if we don’t invest in transit because autonomous cars are coming? That’s almost like 5-year gap between UW Station and U-District station, or the gap between when Link started and buses leave the tunnel, except it’s a lot longer.
“The technology will improve. Cars will learn to optimize their following distance (but that means that they will be 4 seconds apart, which drivers today are too impatient to do, so that means less road space available). Car computers will learn to talk to each other (hopefully better than Microsoft and Apple), and software bugs will be worked out so that computers don’t “crash” while they are in the driver seat. Everyone will want one, and be able to afford them. If you believe all that, I’ve got a bridge to sell you.”
Insulting the person who disagrees with you, while offering no other argument or evidence against their beliefs, is about the weakest argument you can make, and juvenile, at that. You undermine your own authority by doing so little to back up your claims.
Claiming that technology will not advance to be better, cheaper, and widely used is a pretty bold argument. Equally bold is claiming that you know how that technology becoming widespread will or will not change the world.
I’d bet on tech advancing and becoming insespensible – that’s what it does in whatever areas it touches. I’d bet that driverless cars will make car sharing cheaper by getting rid of the driver, and therefore more appealing and prevalent.
Like with the advent of the car, we don’t know how it will change our cities and landscape. Will driverless cars lead to two hour napping and laptop-using commutes? To traffic jams as cars circle rather than paying to park? Will we get rid of parking and have cars drive home, doubling vehicle miles traveled and creating a second morning and evening peak with the opposite flow direction, but nobody in the vehicles? Or will we get rid of parking, usually transport a person in the other direction, and make our cities denser? Will autonomous cars lead to car-, van-, and minibus-pooling, increasing the density and walkability of urban places? Will that lead to increased transit use?
I think that to sweep these questions aside is irresponsible and makes for a boring article. I agree that fundamentally, large fixed route vehicles on dedicated ROW have the most passenger capacity for the space they use, and therefore are a good investment in a dense city. I agree that we don’t want to sacrifice the pedestrian environment for vehicle speeds. But I don’t think we know yet if autonomous vehicles will compete primarily with the bus or the private car, if we will prefer privacy or saving money by carpooling (especially in small vehicles), if mid to low density neighborhoods will be served more efficiently by on-demand vans with a half-dozen passengers or by the city bus, and if the former, if it will be good enough to out-compete driving alone for most people, and therefore be a CO2 and congestion reducer.
I’m in no way saying the anti-transit people are right. Rail provides incredible capacity and reliability, and is excellent insurance for an uncertain future. I’m just saying that this article seems to be dismissing the importance of the smart phone during the era of the blackberry – when we could first see that it was coming, but didn’t know what it looked like yet.
Sorry, I went overboard. That paragraph just really irked me.
Technology will advance, but technology can not change physics.
Safe following distance is severely limited by physics, such as brake shoe friction and tire on road friction. Advances in computing power alone won’t change those limits.
Yes, that’s why grade-separated mass transit makes sense: it has proven, huge capacity. The recent NYTimes article on the (awesome) problem of Manhattan sidewalks running out of capacity shows that rail capacity can support about as much density as we really want. We’re a long way from having driverless vehicles have that kind of capacity.
However, I disagree on the safe following distance question. The other vehicles on the road are limited by the same laws of physics as yours; it isn’t possible for the car in front of you to come to a stop faster than your car can come to a stop.
In a freeway-type environment, where every object is either fixed or talking to eachother and has the same braking limits, you could have a whole chain of cars simultaneously hit the brakes with the same force, and all come to a stop without changing the distance between any of them. The limits on following distance here would be a question of how quickly a problem could develop and how much variety you have in the physics due to things like a wet patch of road.
This is very different from current conditions, where safe following distance is governed primarily by human reaction time. You need the same gaps as above, plus a gap between each car long enough for humans to realize they need to hit the brakes. The signal of problem (seeing somebody’s brake lights) goes from one person to the next at a speed governed by reaction time, and that’s why we need large following distances. (that’s why it’s even possible for race car drivers, trained to have lightening quick reactions, to drive bumper to bumper at high speeds. Computers, of course, have much faster reactions than that, and we don’t need to go 200 mph on the highway, either).
Of course pedestrians make this all much more difficult, so the gains in urban areas are lower. And it all requires not just tech development, but the development of standards and the regulatory environment, and the elimination of human drivers on relevant sections of roadway. It’s gonna take a while.
But development of safe driverless cars is primarily limited by the technological development. If we’re going to try to predict the future and plan for that future, we should probably look at what will happen when there are driverless cars, but not yet what will happen when there aren’t human drivers.
One other thing. I’m a transit advocate primarily as a function of being an urbanist, rather than the other way around. I think suburbs are an environmental, cultural, and aesthetic disaster, whereas cities provide a vibrant life with a much smaller carbon footprint.
If you see transit as the good, and density as the means, then driverless cars competing with transit is a problem.
If you see density as the good and transit as the means, then competition with transit isn’t such a bad thing. In that perspective, the ability to live way far out in the ‘burbs and just take a nap or check your email on the way to work scares the heck out of me. On the other hand, the potential to use space more effectively, by having cars be useful for most of the day, rather than waiting for their driver, and by making on-the-fly, point-to-point carpools, is hugely exciting. However we get to more efficient use of road space, it means more stuff within walking distance, more space for sidewalks, more dense living, more mobility to more stuff for more people.
Or maybe more suburban sprawl. Or maybe both. We’ll see.
under what scenario does this row of cars at close following distance come to a safe stop?
Suppose a tree falls into the road. The fist car can’t stop in time and hits the tree. This also means the 20 cars behind can’t stop in time and hit the car in front of them.
Or maybe a deer runs across the road.
Or maybe something falls off a semi truck.
Lots of unpredictable things happen out there that can not be dealt with by simply tightening up the following distance and having everything talk to each other.
There’s also the little issue of changing lanes and merging. To really work well, the following distance has to be set to allow enough space for that to happen.
One thing that writers, researchers, and other advocates for driverless cars often forget is the “meantime”. Perhaps one day, Seattle will be a utopia of networked autonomous vehicles, never crashing, never clogging, never in the way. But how far in the future is that? 10, 20, 30 years? In the meantime, the streets and highways will be filled with some percentage of human drivers. Even if driverless cars could solve every conceivable urban problem, they would still have to share the road with imperfect humans for multiple decades. During this transitionary era, it might be hard to see the benefits of driverless cars.
But the way, excellent timing for this article:
Another Tesla crash:
http://www.bbc.com/news/technology-36783345
Supposedly the car had asked for manual control to be resumed but the driver ignored this.
Rent-by-the-ride (Uber, etc.) driverless cars won’t work for families with young children. They require right-sized car seats for children up to 8 – 10 years old.
Several of the carless families near me appear to own car seats so that if someone does give them a ride someplace they have a solution to that.
I suspect that the most useful autonomous vehicles will be of the ‘last mile’ sort, and think more ‘glorified golf cart’ than sedan or bus. What if the #5 and E Rapid Ride buses could stop every 5 and 10 blocks, and be fed by small vehicles traveling at not more than 15-20 mph, getting riders to the stop. Other such small vehicles should be serving the downtown hills. I like and do walk them but people with children and elders (actually I am there now) often need the assistance.
My futuristic dream, use a smart phone to get from Greenwood to down town, it tells when the golf cart will be within 2 (?) blocks of you, and when there will be a space to get on the bus. (E bus is often too loaded to pick up more passengers).
perhaps driverless cars will make park-and-ride obsolete.