« bus maps for developing-world cities | Main | a leading bureaucrat on the need to take more risks »

TrackBack

TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00d83454714d69e2017ee8007a44970d

Listed below are links to weblogs that reference a transit manager on driverless cars (guest post by ron kilcoyne):

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

Laurence Aurbach

Good post, Ron. Another factor I haven't seen discussed is the cost of the system. If the price tag turns out to be a substantial percentage of the total vehicle cost, then self-driving systems may be more cost effective for higher-occupancy vehicles. The financial considerations might be similar in some ways to the current automated transport industry. Only two or three personal rapid transit systems have been built, using small, car-like vehicles. By contrast, 90 or more people mover systems have been built, using vehicles similar to regular trams and trainsets. Logistics and economics have favored the high-occupancy solutions.

It may also turn out that self-driving technology is safer operating on dedicated routes/lanes. If so, the condition of limited available road space would tend to favor buses, vans, and other high-occupancy vehicles.

Alex Forrest

Ron points out that driverless cars make reading and doing work possible while commuting, as transit does, but it's important to remember that this is an advantage specific to long commutes. Whatever vehicle people are in, I'm skeptical that they'll break out their laptops for a 5-10 minute casual trip.

bradtem.myopenid.com

You're right that our predictions of the future will have flaws, and people will do things we don't expect. However, I feel confident in predicting that the way we think about urban transportation, both by cars and by transit, is in for a huge change, and the only sure mistake is to assume it's going to be just like today.

I have a number of articles on these topics for you to consider at robocars.com but let me open up a few issues for you:

Sprawl. Sprawl has many elements, but two important ones are low density housing (which is probably not going away) and automobile-oriented commercial zones -- buildings set back from roads, surrounded by parking, rarely walked to. As the robocars come, both parking themselves more efficiently and just dropping people off and going off to do other taxi jobs (car share) we will see the parking lots slowly emptying out, even at Christmas. The landowners will be rational and say, "I need to do something more efficient with that land." I have some forecasts of this in my article on Robocar Oriented Development at http://robocars.com/rod.html

You're right to fear that transit investment will drop in the robocar world. Transit turns out to be efficient at rush hour, but less efficient than cars off-peak, at least in the USA, for a total average that's not very good. The rational and green thing to do is to have properly designed transit (including automatic vanpools as a prime method) at rush hour when road congestion is an issue and load is high, and leave things to robotic taxis off-peak, especially efficient "city car" style ones for trips that never touch the highway. However, that's not like today's transit planning at all.

It's time for research in the urban planning community. We can't be sure which way the future will go but we should investigate the various probable paths so we can choose wisely. People can only have their behaviour bent so much. They will get what they truly want but we can adjust things a little if it turns out to give them that in a better way. They crave private, door to door transportation (even at a high price, but especially at a low price) without parking hassles and the ability to read while riding. The robocar is going to give them that.

Liked off the rod article you will find some other questions I want urban planners to discuss.

EN57

There is no solution to urban traffic congestion with robocars. Just more people being stuck for more hours, days, years in traffic. Except that people won't be in control of the vehicles - which will make it just as frustrating as riding transit. Real, live people going about their business will need to yield to empty robocars circulating the streets trying to find parking.

How do robocars work when 80,000 people simultaneously input "Downtown" as their destination at 8am each morning? (remembering that people are free to choose their own destinations and times just as they desire in accordance with their human nature – thanks Matthew Newton). How do thousands and thousands of robocars marshall themselves in the CBD at 4.45pm in preparation for the afternoon deluge of passengers as required by demand? If people have to modify their behaviour to make this system work, where’s the freedom? and how is this any better than what we have at the moment?

Andre Lot

I fell an enormous resentment and even outright prejudice against self-driven cars because they can get in the way of other urban-promoting agendas.

Let's all chill out and think outside the box.

For self-driven and non-driver-assisted cars being not able to be 100% fail-proof, I don't think that will be so much of an obstacle for its widespread adoption. We have many industrial systems (think of escalators, elevators, automated gates) that aren't perfect either. I think once the public realizes a 99.9995% reliable self-driven car is much safer than an human-driven traditional one, support will drastically increase. In the past, fear of automated systems all hold back a little things like non-assisted elevators or sensor-based secure gates, but no longer.

Nonetheless, there is a huge potential for a world full of self-driven cars. For a starter, park and ride operations become much easier to manage. You don't need thousands of people driving on the same freeway peak time, but with cars the challenge of last-mile transportation is eased.

Self-driven cars can also provide much needed convenience. I envision things like serious - but non-emergency - health crisis where you jump on a self-driven car that takes you to the hospital. It is also possible to envision almost-instant delivery of online purchases if you want to, without the need for a courier to drive to your place, just one to put a package on a self-driven small van that will park near your house.

The whole parking discussion is also facilitated, because you can then park a lot of self-driven cars on multi-stack places where they don't interfere with pedestrian uses or take up a large chunk of land while idling for hours.

Now, I understand why the planning community is so leery of them. It threatens a basic mantra we read nowadays, that you need to build cities for transit as much as the opposite. But I think a more holistic approach is needed. Self-driven cars are not the "enemy" (I read in other blog some activist suggesting a "10-year moratorium" until they can "regroup to discuss its impact").

Alex Forrest

Andre, I wouldn't expect self-driven cars to be an "enemy" of transit or of cities in general. On the contrary, they provide private transportation without requiring as much road or parking space as ordinary cars. To the extent that they promote pedestrian mobility, self-driven cars can be an excellent asset to the city.

What is concerning, though, is the idea that self-driving cars--or really any mode of transportation--can serve all travel needs singlehanded. The fact that people usually contrast self-driving cars with transit, rather than with conventional cars, suggests that this line of thinking is prevalent. The same thought process is what led cities to abandon their pedestrian spaces in favor of roads and parking lots: all other modes of travel were viewed as outdated and irrelevant.

The panic on the part of some blogging communities suggests that many people aren't very clear on what makes transit so useful in the first place--not hands-free transportation, but maximized passenger throughput per foot of ROW width. The degree to which transit nodes can concentrate pedestrian activity can't be replicated by personal vehicles: no matter how cleverly engineered, personal vehicles consume too much space per passenger to replace a dedicated transit corridor, regardless of mode. Jarrett has made this point a few times already, and it's still true--geometry matters!

Robert Wightman

Has anyone calculated the number of self driven cars that will be needed to get everyone to work at the same time, or in the same time period? There is still the problem of parking these cars or do the continue to circulate around the streets empty burning fuel while looking for passengers.

As several people have commented the numbers do not add up. Making the cars driverless is not going to allow any more cars to get on the current road system than happens now. If you live in an area with all one way commuting, into the city in the morning and out in the afternoon, then you might be able to increase through put with reversible lanes but that happens with human drivers also.

Aerial gondolas are the answer! They just keep going around and around in long loops; just like the call for new space age systems to make public transport obsolete.

Matt

"We are not capable of designing 100% foolproof systems."

But we are very capable of designing systems orders of magnitude more foolproof than the damn fool humans behind the wheel.

Matt

"Whatever vehicle people are in, I'm skeptical that they'll break out their laptops for a 5-10 minute casual trip."

People break out their phones to text when they're stopped at a light. I'm sure they'll find something to do for 5-10 minutes.

"Making the cars driverless is not going to allow any more cars to get on the current road system than happens now."

Driverless cars can drive closer (both side to side and front and back) safely than humans can (due to reaction times and predictable behavior). That fits more vehicles in a given space. They also won't rubberneck and cause congestion that way.

bradtem.myopenid.com

There are a lot of things you can do, depending on the city, for the situation of "everybody wants to go downtown" in the morning, and I have outlined a number in my essays. While we're some decades away from it, grid-based cities have absolutely huge capacity if you can do things like seriously reduce accidents, time lights and traffic flow, clear away all street parked cars, redirect 3/4 of the streets one-way in the rush direction, load balance over the entire grid of streets, encourage smaller vehicles, reduce headways and do congestion charging to meter usage.

If you start combining unscheduled vans and buses which use small robotic taxis to feed them coming in and to distribute their riders at the destination zone, the capacity of road lanes filled with such buses can be absolutely astounding, beyond anything conceived with today's transit modes.

But instead, let's look at the question at a higher level. Cities are moving away from pure CBD style to a more polycentric style. That means that you don't have everybody going downtown. This is a problem that limits the ridership of many transit systems, which only function well to and from the CBD and do poorly at the majority of trips which today do not involve the CBD.

We're also seeing more variation in work hours (in response to the commute being a bitch if you work 9 to 5.) With full metering you could simply say, "Sorry the capacity on these congested roads is X cars/hour, so we are not allowing more than X cars, period." Whether it's the top bidding X cars, or X lottery winners or people lining up on highway onramps as they do today, when people become clear they are not allocated road use, then they do something else -- move to transit, move their house, move their job, or most likely, move their hours.

Computers can do more than drive the cars and make their movements more regular. If we regulate traffic we can eventually move past the idea that more people try to use a road than it has capacity to handle. Once you reduce the accidents and the irrational slowdowns of human drivers, the roads stay at their capacity most of the time.

I am fond of using markets here, but some oppose congestion charges because they can make the prime road space only for the wealthy. If that's the case there are other methods I have outlined in other essays at robocars.com

JohnHupp

I have gotten the impression that "electronically chauffeured vehicles" will have the biggest impact on limited-access highways. First and foremost, backup control systems will have a positive effect on road safety, continuing a decades-long trend of decreasing rates of traffic fatalities. Fewer accidents means fewer SigAlerts means fewer red spots on Google Maps.

Even with no traffic, there are plenty of insane commutes here in Southern California. People can have 50-60 mile commutes if they're really that masochistic. I don't think driverless cars will have a perceptible effect on people's desire for shorter commutes, which usually ultimately means transit.

Driverless cars seem like they would be most competitive short-ish long-haul drives like Los Angeles to Bakersfield, which is to say they would be competing with regional rail, not urban mass transit. And given how long we are going to have to wait for the California High Speed Rail, I don't think stop-gap innovation at the scale is a bad thing.

Alex Forrest

"If you start combining unscheduled vans and buses which use small robotic taxis to feed them coming in and to distribute their riders at the destination zone, the capacity of road lanes filled with such buses can be absolutely astounding, beyond anything conceived with today's transit modes."

On the topic of capacity:

At 60 mph, with human drivers, a single lane of highway can typically process 2,000 cars per hour (one every 1.8 seconds), which amounts to 2,000 people per hour if we assume 1 person per car.

Suppose the improvements associated with automation allow a staggering tenfold increase in capacity per lane: now you could move 20,000 people per hour! How does transit compare?

Tokyo's Yamanote Line runs 11-car trains every three minutes. The first and last carriages carry 143 people, the other nine carry 162. This means that each train is designed to carry 1,744 people (during rush hours, the *actual* passenger counts get much higher). At existing frequencies, this amounts to 34,880 passengers per hour per track.

So just to recap: 35,000 people per track vs 2,000 people per lane today, or perhaps 20,000 people per lane in the future. There are real advantages and opportunities in self-driving cars, but they aren't in the same league as transit when it comes to capacity.

Richard

The real transformation will be caused by the move away from the ownership model for transportation to a service model. People who don't own a car chose from a varity of options to get around. The problem with owning a car is that it is easy and cheap to chose it for a trip even when it is not the best or most efficient option for that trip. For shorter trips, they will likely chose walking or cycling. Many people love to walk and cycle and often do so even if they have a car. Auto taxis will not change that. They have the potential to make walking and cycling much safer and more pleasant. We need to start now on creating the laws and policies that govern their use to ensure they benefit cycling and walking.

Auto taxis will be best for short to medium length trips. The cost of use will likely be enough to discourage people from using them for long trips or too often for that matter. I would imagine the cost structure would be similar to that of car sharing today. Initially, it would be higher do to the higher vehicle costs. People who have long commutes will probably be better off owning their own vehicle which increases their transportation costs dramatically. Due to this and the pricing of trips by time or km, like walking, transit and cycling, auto taxis will likely encourage and support compact walkable unbanked environments.

Robert Wightman

bradtem.myopenid.com says:

"There are a lot of things you can do, depending on the city, for the situation of "everybody wants to go downtown" in the morning, and I have outlined a number in my essays. While we're some decades away from it, grid-based cities have absolutely huge capacity if you can do things like seriously reduce accidents, time lights and traffic flow, clear away all street parked cars, redirect 3/4 of the streets one-way in the rush direction, load balance over the entire grid of streets, encourage smaller vehicles, reduce headways and do congestion charging to meter usage."

As Jarrett says making streets one way is not good for transit, maybe cars but definitely not transit. Making 3/4 of the roads one way inbound in the a. m. then out bound in the p. m. assumes everyone wants to go to the CBD. What happens if you live in a city where the reverse flow is almost equal to the peak direction?

Load balancing over the entire grid would be good if the people actually wanted to go in an evenly distributed fashion but unfortunately many want to go to a small area so there is no road capacity for the required number of cars regardless of how small they are.

"But instead, let's look at the question at a higher level. Cities are moving away from pure CBD style to a more polycentric style. That means that you don't have everybody going downtown. This is a problem that limits the ridership of many transit systems, which only function well to and from the CBD and do poorly at the majority of trips which today do not involve the CBD."

This is true if your transit system runs in a radial mode but if it runs in a true grid then it is much easier for people to get to the polycentric nodes.There are transit systems that do operate in a true grid. The other requirement is for free and convenient connections (transfers.)

I agree with most of your other comments about congestion charging and staggered work hours but what you suggest in the passages i referred to would not work in many cities.

Jase

It wasn't long ago I was a big believer in driverless cars.
I thought Jarrett's dubiousness was stupid.
But I've changed my mind.
I still think driverless cars would be amazing. But I don't think they'll happen.
Our problem is this: we're applying a heuristic to technology change in this space that we've got from observing other areas of technology.
We see the speed at which the internet, smartphones and communications develop, and assume that can be applied to sensors and spatial computation.
But there are two very different kinds of invention.
The first is coming up with a wholly new technology, and seeing if and how humans can use it.
Things like Twitter, the wheel, microwaves are all examples of this.
When the invention is good, it spreads like the proverbial clap.
The second type of invention is trying to solve an extant problem. Things like setting out to cure cancer, inventing a robot vacuum cleaner, making a jetpack so people can fly, making a better battery.
In each case, we've made major breakthroughs, sure, but the technology moves much more slowly than we'd like, and much more slowly than in other fields.
Solving a problem to human satisfaction means achieving success on a whole range of fronts, technical, cost, safety, user interface, etc..
At the 1984 Olympics, we might have believed individual jetpacks were around the corner. They are not. We are probably in a similar position now with driverless cars.

Looking at it another way, we have a 90/10 problem. What seems like 90 percent of the work will get done in 10 per cent of the time. Right when we seem to be on the brink of success, some little bug in the technology or legal environment of driverless cars will prove so wicked and intransigent that they will never become mainstream.
It's a bit like those robot vacuum cleaners. To get them to work at a level that's not completely frustrating, you need to buy one that costs $400. Even then it still gets stuck or does dumb things from time to time.

I'm sad about this, because I hate car crashes, traffic and wasted time driving, but it's my new position on driverless cars.

Richard

@Jase

Our current oil, ownership and driver based transportation system is so inefficient in terms of energy, land, resources and time that is has to change sooner or later. There is no other option. Almost all the energy used for passenger transportation by car is wasted. The energy used is much more than required to move the weight of the passenger. With such huge economic advantages to movement of people and goods, it will happen sooner or later.

I expect it will happen first in China where the are building whole new Cities from scratch. When building a whole new city, they can do it in a way to better accommodate automated vehicles. They also have a great high speed rail system thus cars aren't needed for long high speed trips. People will be able to use small very efficient light weight automated vehicles for short trips around cities.

China also spends billions sudsidizing expensive imported oil so the have a huge incentive.

bradtem.myopenid.com

To answer a few questions:

a) A single lane of freeway with robotic 40 person buses and 2 second headway would move 72,000 people per hour (seated) dwarfing the 35,000 (mostly standing) people on the heavy rail. At one second headway it's double. Not that we would do this, we don't need to. But the nice thing about a robotic, unscheduled transit system is you use the capacity you have. If you only have 3,000 people per hour to move, you can do it with a lane of road (average is 1.47 people/car.) If you have more, and you are METERING the road, you start saying, "Hey, there are 40 people all going from exit A to exit B -- let's offer them a bus, and raise the congestion price for the cars until they switch." And you keep doing that.

b) But, as I noted, you can load balance over a whole grid in many cities. You ask, why would somebody go out of their way to use a grid street 5 blocks west? Well, because there's metering and the direct street is full. The non-full street that's a bit out of the way is better than a clogged street, if you're even allowing the street to clog.

c) Redirecting 3/4 streets is something you do when the flow is mostly downtown. If you get to a world where you can redirect streets in a dynamic fashion, and drivers are directed by their cell phones to follow the right path, you redirect what you need, not more, not less. Today there are lots of redirected lanes on highways and bridges, but we only do it when we can make separated lanes if possible. Once you have a smarter computerized system you can do a lot more. Redirecting to one-way is safe for humans -- robots could even handle lane by lane redirection. It does require every driver have a smartphone that is tracking the street redirections and listening to it. That's not too far in the future if we want it.

d) I challenge the statement about the success of grid transit systems. Riders hate transfers, at least in the US. (Japanese and Swiss trains are a different story.) They hate them with good reason. Especially on less frequent non-CBD lines.

People going between two points in a city not via the CBD want to go directly, door to door, and they overwhelminigly choose cars for this even when they might ride the train into the CBD. This is the curse of the transit line though -- because once people see even a modest number of trips they need to take that the transit will not serve well, they cave in and buy a car. And having bought it, they use it even if the transit might have done a decent job.

Transit Riding Transit Planner

"a) A single lane of freeway with robotic 40 person buses and 2 second headway would move 72,000 people per hour (seated) dwarfing the 35,000 (mostly standing) people on the heavy rail."

Except for that slightly inconvenient part where passengers have to somehow board and alight in less than two seconds...

Marc

Why oh why is Jevons Paradox repeatedly ignored in all the SDV discussions?

Any attempt at increasing travel efficiency/convenience WILL result in more travel and more congestion. If SDVs allow for increased vehicle loads on any given segment of a highway, the width of the highway will NOT be reduced as suggested above...

"If we can accommodate twice as many cars on a freeway with self driving cars lets shrink the freeway in half."

...No, the highway will retains its current width, except all the lanes will eventually be filled with SDVs.

Furthermore, SDVs will continue to require ever-more parking lots/garages: Empty SDVs will still need to retire to nearby lots or retreat to distant ones which will - surprise! - increase congestion because more and more vehicles will be making trips that weren't previously necessary. Right now we have a lot of conventional cars making trips with one person in them and the best idea for the future is: to have empty AND single-occupancy SDVs scurrying around?

History shows us that, no matter the mode, the more easy, cheap, convenient, and efficient you make travel, the more travel - and thus congestion - you will get. There are no exceptions.

But IMO the silver lining is: the SDV debate is just idle speculation. Maybe there'll be some boutique SDVs in the hands of the rich, and maybe even some multiple-occupancy SDVs replacing buses and whatnot, but we'll probably discover that the technology just can't scale up, like the IVHS's of the 90s and the PRTs before then. And this is apart from the issue of future fuels: assuming that gas continues its "three steps forward, one step back" upward march, that we already (disappointingly) realized that ethanol and hydrogen can't scale up, that our "vast" supply of shale gas is actually concentrated in a limited number of sweet spots, and that we haven't been able to scale up EVs for over 100 years, what fuel would all the SDVs 50-100 years from now run on?

Mat

Marc, SDVs as you can call them could easily run elecric cars at scale, using the technology that currently exists. That's what makes them exciting.

Google is absolutely determined to ram this through, the founders are extremely excited about it and view it as their best chance to have a positive impact on the world.

Not only that but to every fleet manager and logistics manager in the entire world it is going to make enormous sense, EVEN if they have to keep the driver behind the wheel.

This is definitely coming.

Marc

Mat,

What existing technology? Gas-electric hybrids remain expensive curiosities, and EVs are even more exotic.

Right now there is a lot of hype and hysteria over NOTHING. There are only a handful of SDVs in testing. If that's enough to conclude that mass SDV use is right around the corner, then by this logic I can conclude that conventional EVs are even more of a reality than SDVs. After all, there actually are even more EVs in testing than there are SDVs, and even a handful of EVs in regular use among ordinary consumers. Yet far fewer people are deluded by the notion that EVs are right around the corner!

EVs and hybrids can't scale up - we've tried for 100+ years already. So far SDVs haven't scaled up either. Maybe this will indeed happen someday, but I'll wait till it actually DOES before dedicating so much time, energy, and transit planners' attention to hypotheticals. Let's let the technology incrementally, privately mature by itself - like canals, railroads, conventional cars, and airplanes did - before we redirect public resources to accommodating it. Transit planners still have a lot to do before (if) the technology ever scales up, and it annoys me how this silly hypothetical is repeatedly served up as a distraction.

bentleychauffeur

There are a lot of things that you have explained.The driver based transportation system is so inefficient in terms of resources.But London, is an important city filled with great sights that must definitely be seen.

Verify your Comment

Previewing your Comment

This is only a preview. Your comment has not yet been posted.

Working...
Your comment could not be posted. Error type:
Your comment has been posted. Post another comment

The letters and numbers you entered did not match the image. Please try again.

As a final step before posting your comment, enter the letters and numbers you see in the image below. This prevents automated programs from posting comments.

Having trouble reading this image? View an alternate.

Working...

Post a comment

the firm

Jarrett is now in ...

Related Posts Plugin for WordPress, Blogger...
Related Posts Plugin for WordPress, Blogger...