Fact-Checking the "PRT Boondoggle" Blog
A project of the PRT NewsCenter

Saturday, June 14, 2008

"None shall pass!"

It's really fascinating to see where Thermidor the Minnesota anti-PRT propagandist chooses to plant his flag. The latest oddity just went down at Wikipedia.

He talks a big game: "The new tables of would-be PRT vendors are full of errors," he wrote dramatically about lists of PRT projects and their status (some of which are Active and/or Under Construction!)

And yet his criticisms that might be substantive are all knocked down by editors Atren and Skybum:

  • 2getthere is an independent company, not in bankruptcy
  • Taxi 2000 is active
  • 'Levacar' and 'Carveyor' are not PRT
  • Also not PRT are 'Hallitubes' and 'PRT Project' (despite the name)

"I'll bite yer legs off!"
He keeps trying to include odd or failed transportation concepts within the PRT definition hoping to make PRT look odd or failed.

But like the Black Knight from Monty Python and the Holy Grail, he fights on until he's been chopped to pieces, and all he can do his threaten to bite your legs off. In this case, all he's left to argue with is whether the Taxi 2000 Corporation is located in Minneapolis, or its suburb Fridley.

Even though it's like arguing whether something is in Brooklyn or New York City. Indeed, that would do a lot to prove that PRT is a right-wing anti-transit scam. Way to go, Humidor, stick to your guns!

If you see one where people are swimming, shout "look out, there are Ken Avidors!"


gary said...

One correction...prtproject is a PRT concept, if only a concept. Why do you see it otherwise?


Mr_Grant said...

Well, perhaps it is unfair of me keep PRT Project lumped in with Hallitubes, because your idea is neither odd nor failed. However I should point out that I didn't do the original lumping. My structure was in response to He Who Shall Not Be Acknowledged By Name.

As we've discussed previously though, it has one foot outside the PRT definition because of the lack of consistent grade separation, and house to house service is more in the realm of a 'car replacement' than a transit system.

gary said...

I suppose you can call it whatever you like. I actually see my proposal as taking traditional airport-bound PRT to it's full potential. By suggesting that PRT will ultimately replace the automobile rather than just supplement it, I'm actually the biggest fan in PRT around!


A Transportation Enthusiast said...

Gary, there is a big difference between targetting PRT as an eventual replacement for the automobile, and your approach, which is to require the elimination of the automobile. The fact that PRT Project is incompatible with the automobile would make it a non-starter in any situation except for ground-up projects like Masdar.

But there is another fundamental problem with your approach: grade separation. The grade separation requirement of PRT addresses a much more difficult problem than just segregation from autos - there is also the more pressing problem of segregation from pedestrians and bicyclists. For PRT Project to work at grade, it must demonstrate that it can reliably avoid a child running into the street to retrieve a ball, and by "reliably" I mean "with 100% certainty". As far as I know, such systems do not currently exist and are decades away from being reliable enough to deposit on our streets.

Therefore, if PRT Project were allowed to run at all in cities, it would need to be a very low speed system that would react very conservatively to any obstacle - such a system would be capacity-crippled and a poor replacement for existing automobiles.

In other words, there are very good reasons why PRT requires grade separation, and it is this fact that disqualifies PRT Project from being true PRT.

Apart from that issue, there seems to be very little documentation about how PRT Project would work. It seems to be a concept that is even less developed than even the other undeveloped PRT concepts (e.g. SkyTran, which at least has documented its design in detail).

So basically, PRT Project is an interesting concept that has few documented details and seems to gloss over a fundamental departure from pure PRT - grade separation - without giving any detail as to how such a system could work in the real world. For these reasons, I think it fails the minimum standard for inclusion.

Please don't interpret this as hostility toward the concept; I admire all efforts to improve transit in cities. But until your concept is more developed and those questions are reliably addressed, it remains in the realm of "PRT-like concept".

gary said...


I certainly agree that transition is *THE* issue with PRTProject, but still not impossible. I have an entire page on the website dedicated to the issue - please check it out if you haven't already. But I completely disagree with your grade separation concern. Obviously it needs to take small children into account, but a combination of image processing, modest fences and appropriate speeds in neighborhoods will work. We currently have people speeding down residential streets in hummers while talking on cell phones. Are you seriously of the opinion that an automated system cannot complete favorably with this? If that's your believe, how many years away do you think we are? Or do think machine will NEVER surpass human drivers? Please clarify as this seems to be your primary argument against replacing the automobile.

BTW, your goal of 100% reliability is just now how the real world works. All technology...airplanes, trains, etc is designed around acceptable loss of life. Someone dies in this country every ten minutes due to an automobile accident. If zero loss of life is the requirement, no system can ever be built.

You mentioned that I have very little documentation, in particular as compared to SkyTran, etc. You are absolutely correct. I have nothing more than the website. I'm just an individual with expertise in computers, plus a degree in general engineering. No funding, prototypes, or anything. In other words I haven't quit my day job.

Btw, I very much appreciate your feedback and do not at all interpret your comments as hostile. To the contrary, I'm very grateful for your help.


Mr_Grant said...


I would add to ATE's observations my own recommendation that you think about the timeline of your program in relation to competing concepts. By that I mean: first, what we'll term 'classic' PRT -- finally -- starting to be implemented now. Second, dual mode, which likely has a number of years of r&d to go and is acknowledged to be a follow-on to PRT. Dual mode thinker Francis Reynolds has written in terms of another 50 years before it is popularized. Since your's is basically a pervasive application of classic PRT, we have to place its adoption in the middle, timewise. Thus, you are left with a relatively narrow window in which your technology would be used.

Next, I invite you to consider what the dominant paradigm is doing: ways to perpetuate the self-drive car powered by electric or alt-fuels. In addition, it is working on the very fail-safe automatic safety systems that could kick in to stop from hitting a pedestrian, cyclist or animal. After that it is a short step to computer-driven all the time, a la "Minority Report." In other words -- PRTProject without rails.

So my question is -- why not make PRT Project a dual mode project?

gary said...

Mr. Grant,

I have no doubt that the future will see the complete elimination of human drivers in all the remote locations. Today a very small number of people are capable of operating a bulldozer. Or knitting a sweater. Or churning their own butter. In the future a similar small percentage of people will know how to drive "manually", nor will have any need for that knowledge. The world is only getting smaller. There are only so many places left to develop. Nor should we continue to develop. Ultimately we will need less people, not more roads. So the idea of "dual mode" makes no more sense to me than requiring every airline passenger to be able to take the wheel.

And using rubber tires on asphalt makes absolutely no sense environmentally. It's less efficient and uses more petroleum. What's the point?

So obviously I don't agree with the dual mode idea. And I certainly don't see how that would fit AFTER what I'm proposing. The most distant future will be 100% automated, no doubt in my mind. After that I suspect we will just "beam" places.


gary said...

(correction to that first sentance...I meant to say "in all but the remote locations".)

gary said...

More food for thought...

"About 70 billion lb of asphalt is used annually in the U.S. alone, and asphalt usage will grow dramatically in Asia during the next 10 years," notes Arthur M. Usmani, chief scientific officer of Usmani Development Co., Indianapolis, in the preface of his book "Asphalt Science and Technology" (New York: Marcel Dekker, 1997).

"Almost all asphalt used today is derived from the bottom of the barrel--that is, the last cut in the petroleum refinery after naphtha, gasoline, kerosene, and other fractions have been removed from crude oil," Usmani tells C&EN. "Very little is produced from other natural sources."

So any supply or environmental issues with petroleum are interconnected with asphalt as well. As I see it the energy debate is just getting started...


A Transportation Enthusiast said...

Gary, I am a software engineer with 15 years experience, and I have worked much of that time in research domains that require a similar kind of artificial intelligence (fuzzy logic, heuristics, whatever you want to call it) that automated at-grade driving would require.

My professional opinion is that we are at least several decades from operating vehicles at road speeds without full separation from pedestrians. I think you vastly underestimate the uniquely human judgement that is a fundamental part of driving. When we drive, we make dozens of exceedingly complex judgements per second, judgements that we are so used to making that we now take them for granted.

Yes, that judgement is impaired in a drunk driver, but there is still no comparison to automated control. Drunk drivers are a menace, but the fact is that even the most stone drunk human can still operate an automobile on open roads vastly better than the most complicated AI algorithms in existence. There's no comparison. Human judgement, even impaired human judgement, is currently not reproduceable in software and will not be for a long time.

As for your claim that it doesn't have to be perfect, well I disagree. Would you step on a plane if you knew the software would fail catastrophically 1% of the time? There are car accidents caused by faulty human judgement every day, and people accept that as part of life, but when you hear about an accident caused by mechanical failure, people are up-in-arms.

Not to mention the legal entanglements of such a system - can you imagine the lawsuits that faulty heuristics would bring?

Also, there is a huge difference between deterministic systems like airplane controls and non-deterministic systems like simulated vision and obstruction detection. The former can be verified for correctness with certainty, and the only variables are the inevitable breakdowns of hardware, which can be dealt with using redundant and fail-safe designs.

For non-deterministic systems, on the other hand, there is no verifiably "correct" operation. These systems employ statistical and fuzzy schemes in order to mimic human judgement and are almost impossible to "verify". The best you can do is test the system exhaustively and measure how much it fails.

Ask yourself why there are not fully automated versions of stuff like airport security screeners, or security systems in general, or speech transcribers, or text transcribers, or language translators. All of these can be automated to a certain extent, but they usually require human intervention because they're so noisy and stumble easily. Now put one of those systems behind the wheel of a 2-ton car and put it on streets with kids on bicycles - do you see the issue? We are decades away from solving it, and even if we did solve it, the regular road network itself would then be drivable, so why institute a track-based PRT system?

The fact remains: PRT requires grade separation. There is no current way around that.

gary said...


Several decades away? That's great news. From your initial posts I was beginning to think you felt it couldn't be done. For the revolution in transportation that's needed, this is the proverbial blink of an eye. Myself I feel it could be accomplished in five years, given a sufficient push. One thing I should clarify here is that, unlike some PRT fans, I'm not of the opinion that we need to travel at high speed to get the job done. Vehicles don't need to travel faster than about 20 mph down residential streets. Acceleration with electric vehicles is fairly good, so the pace can pickup considerably on the "highways". And besides...who's in a hurry? Relax...drink your coffee...take in the scenery. We're talking zero stress transportation here!

> Human judgement, even impaired human judgement, is currently not reproduceable in software and will not be for a long time.

This is a strange statement. You seem to distinguish between what you call "human judgment" and simple control & obstacle avoidance logic. It's all a matter of sensors and relatively basic computing. My own 25+ years of programming embedded systems tells me that this is a very solvable problem. Much of my career was in videogame development, requiring the simulation of the real world and the AI to navigate it. And for the record, humans CANNOT beat videogames unless the programmer allows them to win.

> As for your claim that it doesn't have to be perfect, well I disagree.

The odds are about 1 in 10 million that you are stepping onto a plane doomed to crash, with about 25% of those crashes due to mechanical failure. So I repeat...the system does NOT have to be perfect. Perfection is not a requirement. So you may want to cancel those flight reservations if you still believe that. For more on imperfect airline statistics, see here.

> Also, there is a huge difference between deterministic systems like airplane controls and non-deterministic systems like simulated vision and obstruction detection. The former can be verified for correctness with certainty, and the only variables are the inevitable breakdowns of hardware, which can be dealt with using redundant and fail-safe designs.

There you go again. You can't use terms like "certainty" in engineering. And while airplane and PRT are different, there is no fundamental difference as you describe. They are both mechanical systems, nothing more, nothing less.

Bottom line is that, fuzzy logic notwithstanding, you agree that "we are at least several decades from operating vehicles at road speeds without full separation from pedestrians." So in reality we only differ on the timeline.

> Ask yourself why there are not fully automated versions of stuff like airport security screeners

Do the math. What's cheaper? A fully automated system or low paid immigrant "security" workers?

> or security systems in general,

Many security systems ARE fully automated...they detect intruders and take appropriate action. Maybe I missed your point?

> 2-ton car

No where near. When you remove the bulky batteries and keep the vehicles at a modest size (4 passenger), it's considerably less than a ton.

One last point. If you think elevated platforms are the solution, forget it. WAY too expensive. Impossibly expensive. With elevated platforms we're talking 10 of millions of dollars per mile. And who want to clutter the skyline with rails? Finally it makes absolutely no sense NOT to reuse the existing roadways. The overall concept of PRT is so damn good (as you no doubt agree), it HAS to eventually replace the automobile. Like you said, ground level control IS possible...


A Transportation Enthusiast said...

Gary, I would be happy to be proven wrong on these points, but regardless of whether it's 5 years or 50, I still stand by two things:

1) PRT is defined as being on a "captive guideway" (see this definition from ATRA). Therefore, PRTProject is not considered true PRT because it shares its guideway with pedestrians and bicyclists.

2) PRTProject relies on technology that does not yet exist - namely the technology to operate in mixed mode with humans. With current automated technology it could probably drive no faster than 5mph without a captive guideway. This places it in a much more theoretical category than even a conceptual system like SkyTran (which relies 100% on currently available components).

So regardless of whether or not we agree on the ultimate viability of the concept, the fact remains that it lacks one of the defining features of PRT and is currently a hypothetical system until someone develops a reliable navigation system that can detect human obstacles with an extraordinary degree of accuracy. These two factors make it inappropriate for the PRT article on Wikipedia, which is what this long discussion started with.

But if you believe in your concept, by all means keep working on it. I truly don't believe you will find a satisfactory solution to the non-captive guideway approach, but if you do prove me wrong I will be the first to congratulate you. :-)

gary said...

Whether my concept fits someone's definition of "PRT" or not doesn't really keep me up nights. Nor does exclusion from Wikipedia. I'm perfectly content in developing my own design, based on my understandings of technology. These really are exciting times to live in with respect to energy & transportation. The public has absolutely no idea what PRT means at this point, but that is about to change. The immutable logic of economics and science will allow nothing less. One year ago the public was focused on biofuels, even though the premise flies in the face of basic high school science. Fortunately the only ones left still talking about biofuels are the politicians (always the last to catch on). So yes, the times they are a-changin'.

And while we might not agree on the details, your perspective is greatly appreciated.