Pretty Ridiculous Transit

After the Antiplanner started writing about driverless cars, I received a lot of emails congratulating me for jumping on the PRT bandwagon. I just had to roll my eyes, as I’ve argued since 2003 that driverless cars are the reason why PRT, short for personal rapid transit, will never happen.

First proposed in the 1970s (see these 1976 studies done for Minneapolis-St. Paul and Denver), PRT has attracted a fanatical following. Even though no PRT system has ever been built, they are convinced it is cheap, fast, convenient, energy efficient, and everything else that ordinary transit is not. If you believe everything you read on the web, you would think there are at least a dozen gigantic companies producing PRT lines, including Taxi 2000, ULTra, and many more.

ULTra actually built a 2.4-mile line connecting a parking lot with the terminal at Heathrow Airport. Including 18 vehicles capable of going 25 mph (a bit short of the 180 mph promised in the video), this line cost $41 million, or $17 million per mile. While it may save Heathrow the cost of bus drivers, this is hardly cheap considering freeways cost as little as $2.5 million per lane mile plus right of way (which probably isn’t included in the $41 million Heathrow cost).

The one demonstrator PRT line built in the U.S. is in Morgantown, West Virginia. To work properly, PRT is supposed to be a grid network reaching every corner of town, but the Morgantown example is only one line. Like pretty much every rail transit plan in America, it ended up costing far more than projected, with the price tag rising from initial estimates of about $15 million to $130 million, or about $16 million ($37 million in 2010 dollars) per mile including 71 vehicles capable of going 30 mph. According to users (as reported by an intense PRT skeptic who also happens to be anti-auto), it frequently breaks down.

While ordinary PRT vehicles are captive to the system, the PRT cultists attracted to my driverless cars article focused on a variant known as dual mode. In this system, cars would operate independently on ordinary streets, then somehow join a mass transportation system on major roads. As with PRT, there is a huge cottage industry of web sites devoted to competing versions of dual mode: TriTrack, MegaRail, Rapid Urban Flexible (RUF), the incredibly clumsy Pullway, and a clunky Japanese vehicle that has both rubber tires and steel wheels (which would add thousands of pounds of weight to the vehicle).

Dual-mode advocates are right that only a dual-mode system can combine the flexibility of automobiles with whatever are the supposed advantages of PRT. The problem is that every dual-mode (and PRT) proposal requires a massive investment in new infrastructure. Driverless cars will not.

Sensors such as the lasers indicated within the oval will detect potential obstacles all around driverless cars.

Although some of the original driverless car proposals called for magnets in roads that the cars would sense, auto manufacturers have given up on highway agencies ever installing such magnets. Volkswagen says that enhanced GPS technology can locate a car to within 2 centimeters of where it wants it to be. When combined with sensors such as radar or lasers, each car can independently go where it wants to go without any centralized computer control that is envisioned by most of the PRT proposals.

Supposedly, an advantage of autos (and dual-mode) over ordinary PRT is that the autos are individually owned and so people would take better care of them. But Volkswagen’s projection of autos in 2028 actually envisions that most cars would be shared, and people would just order the kind of car they want when they want it. That’s possible, but I suspect many people will still want to own their own car, especially if, when they are done with it, they can simply tell it to go park itself.

PRT and dual mode have no future. Passenger transportation in the future will increasingly trend toward driverless cars. This means cars will take over the jobs once done by drivers — steering, braking, accelerating — until the driver will be there only in case of an emergency. Although some legislation will be needed to update motor vehicle law, little or no new infrastructure will be required.

Share

15 thoughts on “Pretty Ridiculous Transit

  1. Spokker

    Are there any demonstrations of hundreds of driver-less cars operating together at high speeds yet? I haven’t seen any, but that’s the point where the technology will be closer to reality than science fiction.

  2. OFP2003

    Trains are awesome, the underground stations of the Metro in DC are impressive spaces. Especially if you’ve ever had to dig ditches for a living – then you know the effort behind these edifices. To somehow think a few bucks from each user will pay for one of these systems is nuts!

  3. Frank

    Sweet video. It’s Knight Rider come true! KITT, can you hear me?

    “freeways cost as little as $2.5 million per lane mile”

    As little as? What’s the average?

  4. Scott

    Do people need more reading time, rather than driving?
    Without reaching home, there is the issue of transporting purchases.
    These would be mostly for a small area, with expensive track.

  5. bennett

    Scott,

    I’m not sure “need” is the right word (again). Do people “need” smart phones so they can access the internet every minute of every day no matter where they are? I think that technology doesn’t really make our lives easier per se, but rather make us more productive capitalist. You can get a lot of work done (without risking your and others lives) if your not driving when commuting. You can also use the time for leisure, unless driving in rush hour is leisurely to you.

    That said I’m with Mr. O’Toole on PRT and Spokker on driver-less cars.

  6. C. P. Zilliacus

    The Antiplanner wrote:

    > The one demonstrator PRT line built in the U.S. is
    > in Morgantown, West Virginia. To work properly, PRT
    > is supposed to be a grid network reaching every
    > corner of town, but the Morgantown example is only
    > one line. Like pretty much every rail transit plan
    > in America, it ended up costing far more than
    > projected, with the price tag rising from initial
    > estimates of about $15 million to $130 million, or
    > about $16 million ($37 million in 2010 dollars) per
    > mile including 71 vehicles capable of going 30 mph.
    > According to users (as reported by an intense PRT
    > skeptic who also happens to be anti-auto), it
    > frequently breaks down.

    It is important to mention that Morgantown is the home of the University of West Virginia, and is essentially a college town. As I understand it, the PRT system there serves students, faculty and staff. Much like campus shuttle bus systems do at many institutions of higher learning in other college towns.

    And the PRT system would not be there today without the ability of West Virginia’s longtime Sen. Robert Byrd (D) to procure a lot of federal taxpayer dollars for this project.

  7. C. P. Zilliacus

    Spokker wrote:

    > Are there any demonstrations of hundreds of driver-less
    > cars operating together at high speeds yet? I haven’t
    > seen any, but that’s the point where the technology
    > will be closer to reality than science fiction.

    Back in the 1990’s, the USDOT and GM did such a demonstration (with relatively conventional-looking Buick cars) in the HOV (now HOT) lanes on I-15 (Escondido Freeway) in San Diego County, Calif.

    You can read more in this document (Adobe Acrobat .pdf) from the California Partners for Advanced Transit and Highways programs’s Web site (part of UC Berkeley).

  8. prk166

    @Oded Roth, I’m not sure how limiting the inputs makes PRT “more simple”. More simple than what? More simple than automated cars? If that’s the case I don’t understand how even if it’s more simple, and I’m not so sure its, it matters. With automated cars, one doesn’t need software to manage the entire system. It’s a moot point. Each car only has to manage itself. It only has to pay attention to the area immediately around itself and the route it’s one. Now maybe the routing is what one sees as being more complex. But we already have the routing piece working in every day use. Sure, it’s not in the context of an automated car. But that’s what hundreds of millions of GPS devices are doing. That’s what Google maps does when we ask it for directions. There are high end cars with cruise controls that will automatically adjust speeds to keep their distance from cars in front of them.

    The problem I have with PRT is that we have yet to pull off building any system of any size. In fact, in the case of Denver International Airport when we’ve tried building a large system it’s failed. Denver didn’t attempt building a PRT in the pure sense. But their automated baggage system was for all practical difference the same difference. It just moved bags instead of people on fixed guideways. And those fixed guideways had “the limited number of input block” that supposedly make it more simple to manage and program. The $200 million system was eventually used but it never worked in it’s complete form. It only handled inbound flights for UAl. It didn’t handle any inbound bags that were to be transfered to an outbound flight. It did nearly nothing it was originally built to do. Eventually even UAL gave up on it and shut it down.

    Denver’s system was a complete failure. It’s a well known case for software engineering (e.g. http://www.cis.gsu.edu/~mmoore/CIS3300/handouts/SciAmSept1994.html ). Even if the number of inputs and outputs are limited, it’s still a very complex system. I don’t see any reason why moving people would be any easier. For example, a system to serve the city of Denver would need to be hundreds of miles long and have hundreds if not thousands of inputs. A system 10 times as big as at DIA would be 100 times more complex We’re talking about a system 20, 40, 100 times larger. It would be 200, 400, 1,000 times more complex.

    All the cars in the system would need to be managed automatically. Cars would need to be shifted around the system to meet localized demand (line balancing may sounds simple but the problems it presents like cascading queues are known to be terribly difficult by system engineers). We would need to be able to do all of this timely and safely. The system need to be able to react to problems and even anticipate problems.

    As complex as a self driving car may seem, it doesn’t need to do any of these things. There’s no need for overall system management. Each car manages itself. It’s example what engineers are taught to do when facing a difficult problem, break it down into as small of parts as possible and solve those. Now instead of getting an entire system to work together, move empty cars around, route both efficiently and in ways to keep lines balanced, do it safely for the whole system, the automated car only has to worry about itself. It just has to make sure it follows its own route. It only has to worry about safely interacting with the couple cars that are around it. It only has to keep track and react to it’s immediate surroundings. That is, it’s a much less complex problem to solve than something like PRT for an entire city let alone an entire metropolitan area.

    Sure, automated cars are still a way off. But he technology is getting close. A good test for them will be later this year when a driveless Audi not merely drives but will go up Pikes Peak at race speeds.

    That said, the real problem PRT faces isn’t it’s costs nor that it’s unproven. PRT’s real problem is that if we ever get to the point where we can build an automated system as large and complex as PRT to serve a city like Providence, Philly or even Dallas, in all likeliehood we would have long before that found ourselves able to build and run driverless cars safely. Even if PRT’s operatuinak issues are ever solvable, it’ll be a moot point.

  9. Tad Winiecki

    Randal, I am offended that after all these years of contributing to your blog you left me off your list of PRT and dualmode cultists. I even sent you my PowerPoint presentation on the history and future of personal automated transport, from elevators to spacecraft which showed how driverless cars and PRT fit in.
    While driverless cars can solve some present transport problems there are some they can only solve with high cost and others they can’t solve at all.
    PRT, dualmode, and Evacuated Tube Transportâ„¢ will be able to take a significant fraction of people whare and when they want to go quicker, cheaper, safer and more reliably than competing modes. Of course they can be designed, built and operated badly so they aren’t competitive, but this applies to other modes as well.
    prk166 discussed automated control systems and assumed central control for PRT vehicles and distributed individual control for driverless cars. This is not a reasonable assumption. For example, my Higherway design has three levels of automated control – central routing and billing, local traffic control, and individual vehicle speed, steering, and navigation control. As Oded mentioned, PRT is simpler because there are fewer circumstances and obstacles for PRT vehicles because they are on a guideway.

  10. Pingback: Podcars or Robocars? » The Antiplanner

Leave a Reply