Monday, August 10, 2015

168> Self-Driving Cars and the Urban Wormhole

In our last exciting episode I promised to explore cooperative robotics, but there is an unavoidable side issue that is worth at least one post in itself, one that high-jacks the whole subject. What I am referring to is self-driving cars. It is not just that PRT and automobile automation have such obvious similarities that makes the subject a prerequisite to discussing PRT control. It is also the fact that self-driving cars may actually change the definition and purpose of PRT.

Self-driving cars are not just about following directions and staying on the road. Spinoff features such as automatic braking for collision avoidance are already widely available in everyday production models. Since self-driving features often involve technologies that clearly can enhance safety, there is an “arms race” in this regard.  With communication between vehicles (“I’m slamming on the brakes, so you better, too!”) and awareness of real-time traffic, (such as could be monitored, compiled and reported by the vehicles themselves) it is not hard to see how cars could be made to operate within a sort of “hive mind,” to the benefit of the group. This is basically a PRT operating system and is a wakeup call to self-steering PRT systems like ULTra or ToGethere, whose technologies are getting leapfrogged by this trend. Notably, the self-driving paradigm is proving that a high degree of autonomous control is doable, challenging the centralized control schemes of only a few years back.
Does self-driving auto technology render PRT irrelevant? Do self-driving cars eliminate the same problems that PRT was to solve?

The short answer is no. The question does, however, illustrate how various flavors of PRT have very different challenges in this new technological landscape. What, for example, should the business model be for ULTra in a world transitioning to self-driving taxis that may not need guideways or stations? Of course that may be quite a ways off, and maybe it is Google who should be approaching ULTra. After all, every cent ULTra has made is a cent more than Google has made on its robocars, with Google putting all of their eggs in a business-model-basket that is still a bit of a head-scratcher and is contingent on many  unanswered questions when it comes to safety.

So far Google and the others have seemingly only taken their vehicles out in decent conditions, weather-wise, content to garner impressive numbers of miles without incident. But how do they handle being behind a vehicle that is dropping debris? Or on patchy “black ice?” Fresh snow or street flooding can completely obscure where the road is. Timid response, in these instances, that would prevent lawsuits is also the kind of driving that would snarl traffic. Imagine a car that is afraid to go through ankle-deep water and so just stops! Can a robocar understand when weather conditions are deteriorating too much to enter the freeway or understand the significance of a funnel cloud? Humans usually know when they should stay put or go back. Will Google be able to give that much common sense to their cars? What about morality? Will they know to hit the truck to avoid the woman with the baby carriage?

It has been reported that, at least some states, they cannot dispense with manual controls (such as a steering wheel) nor the human seated so as to operate them. This leads one to wonder under what conditions the authorities would permit such use. Beyond that, what is the profit model that trumps the obvious, added liability issues? Do drivers really want to relinquish all control and do manufacturers really want to sell mechanically stripped down cars with no sex appeal?
A safe starting point would be a special lane or only going at fairly slow speeds. (i.e. a mode similar to current pavement-running PRT systems) After all, many localities allow golf cart type vehicles on public roads, and these need no horns, airbags, seat belts, etc., as they don’t go that fast. While this would probably be allowable, it does little to solve urban congestion and so, by itself, risks irrelevancy.

Interestingly, there are flavors of PRT that have just the opposite problem. SkyTran, for example, is really built more for speed than for serving little stations in every nook and cranny of a typical city. Suspended systems, in general, have the potential to move people from one side of town to the other quite quickly using an inexpensive, minimalistic track, but like all PRT, suffer from the potential problem of not having a sufficient number stations and walk-up customers to create the cash flow to pay-down the system components and still provide a return on investment. It’s the old first and last mile problem. Could self-driving automobiles be the answer to aggregating more PRT passengers at fewer stations? Quite possibly.

Uber has expressed interest in self-driving cars, and car sharing schemes like Zip Car raise interesting questions about ownership. Why garage a vehicle that could be gainfully employed elsewhere while you are not using it? Why own a vehicle at all – especially if there is one parked close by that will come to you when summoned?
Once upon a time, PRT offered a unique combination of benefits that could not be had otherwise. It was automated, elevated, personal, fast, but it was only practical within an interdependent framework of technologies. Over time, developments in sensor, communications and computing gave advantage to a variety of methodologies enabling new designs that emphasized varying missions and business models. The emergence of autonomous cars, it seems, puts a focus on one of the original themes, which is a network of fast, elevated, non-stop corridors that are unencumbered by the gridlock below. This mission is unthreatened by the self-driving car revolution and, indeed, may well benefit from it. PRT, at this point, needs to be considered not as a fleet of automated taxis, but as a network of personally navigable urban wormholes. The automated taxi aspect is now only a means to an end, not the end itself.

Originally the idea was to space PRT stations so that each was reasonably within reach by walking. Now perhaps each station could garage a half-dozen automated taxis, giving such stations much broader reach. This associated service would not have to extend very far to cut the number of stations to a fraction of what would otherwise be required. This scheme works perfectly with car sharing, carpooling, or even private ownership. Even if each fully self-driving “car” can only go a few blocks, with similar vehicles and capabilities at each end of the PRT “wormhole,” the combination could be very synergistic. The timing and payments between the two systems could be integrated, even if under different ownership. On such limited routes where the speed limits are low there is much less reason to require a driver, meaning self-driving cars are clearly ready for this NOW. Where special lanes could be created, ULTra is also ready to go as well.  
Side note: The above combination offers less advantage for the SMART PRT designs shown within this blog because SMART is specifically designed to address the “last mile” problem with vertical capabilities and stationless pickup and drop-off capabilities. Of course any infrastructure, surface or elevated, will undoubtedly face obstacles on certain routes, and so alternatives are always welcome. That being said…

It is important to note that slow, short-haul, self-driving “taxis” also offer synergy with other forms of mass transit, such as light rail, subways, scheduled shuttles and so forth. Even with limited routes, (There might be too much traffic on public roads and no room for a special lane) phone apps could be used to locate the nearest suitable pickup point. Such pickup points could enhance the value of properties that are otherwise inconveniently remote from mass transit. Short hauls to such existing transit hubs seems like a good business model for such a product/service right now and deployment of such FULLY robotic vehicles, even at slow speeds, might serve similar R&D aims as the current, unpaid efforts and would offer a potential opportunity to cement a leadership position in the field while generating revenue.

Self-driving cars, capable of full city and highway use without special lanes or a standby driver are still a long ways off for general use, and the eventual payoff for the producers of such vehicles is questionable. Short range, slower vehicles face no particular obstacle to immediate adoption, however, and everybody from ULTra to Google to Uber to Zip and the various auto makers should jump on this opportunity to become leaders in this transitional space. Meanwhile PRT wannabes need to take note that the current technological backdrop no longer supports slower, station-intensive, short range systems. Instead, PRT companies should concentrate on designs that foster the cheapest, fastest network of elevated track possible, filling that one niche that automation itself cannot address. 
Elevation has always been the answer to surface traffic congestion, but has always been prohibitively expensive (not to mention ugly and in-the-way) when scaled for heavy vehicles, and so gridlock continues. Like fiber-optics compared to copper, some form of light, affordable, high-speed “pipes” for moving people are inevitable, and the first barrier, autonomous yet cooperative automation, is falling away fast. Like that fiber-optic cable, the second barrier is what happens at each end – how data is efficiently transmitted and received. Hopefully self-driving vehicles will help provide comparable, easy-to-implement solutions for either end of the mobility solution that cities need so badly - that Urban Wormhole technology called PRT.


14 comments:

qt said...

Note also the obsolescence of most reasons for "hybrid" PRT. Hanging a car off the rail makes a good bit less sense now...

Dan said...

Dan the Blogger has been up to his "%#^#&" in "*&#%*#"!
Sorry for abandoning my own blog, folks. I've got washed out driveways and a burned up generator up north (almost no power, being off-the-grid) and octagenarians putting their home up for sale down south, with me flying back and forth...

qt, are you referring to heavier, fully road-capable "dual-mode?" Or private ownership?

I am not sure that the case for dual-mode has been reduced so much as the case for inexpensive, high-speed track, has been increased. Certainly the need for privately owned
vehicles becomes less with autonomous control...

qt said...

Sounds like you've got your plate full, all right.

I was referring mostly to the kind of dual mode we dicussing some time back, with the integration of golf carts or Segways or the equivalent as a solution to the last-mile problem. Drive to the station and hook to a bogie. I recall you were looking for an absolute minimum solution to conserve weight, and I was advocating something with a ten-plus-mile range, to make it generally useful enough (for running errands, etc.) to be an attractive purchase.

Only if the golf cart can come pick you up and taxi you to the nearest station, the utility of the privately owned pod/errand-runner isn't nearly as attractive. Which may simplify things for the average urban planner.

As for fully-road-capable dual-mode, I always thought that was more an *inter*-city thing. Having my own car in Dallas without having to drive it all the way from Atlanta is more attractive to me than riding an incredibly fast train and hoping the 300mph derailment doesn't happen this trip. High Speed Rail has never made much sense to me, but I don't think it's an accident that the last passenger train to make money in the States (without D. C. commuters demanding government help) was the Autotrain.

Dan said...

The "#&^&*" continues... qt. Since you last posted, my laptop, my phone, and my car's fuel pump all died up in NH, and back in Texas, I came in to find my internet down, and two days later my step-mom's hot water heater went out. Two octegenarian women without hot water on a Saturday night... not a pretty picture! Even the replacement computer I am typing on now has a problem and is going back. I just hope my old machine gets fixed correctly as I am having Dell send it directly to NH, and heading up there with just a hard-drive. Anyway...
Up north, I drive 20 miles to get groceries and I can't imagine a self-driving car being much use in winter... especially if it has to drive a long way to pick you up! Up there lots of folks have jeeps or pickups with snowplow attachments for their driveways. I can hardly wait to see Google try to teach a pickup how to plow!

Urban transportation, on the other hand, may be on the cusp of diverging once again from the suburban/rural paradigm of private vehicle ownership. (I say "once again" refering to the days of the trolley) There is a critical mass, virtuous cycle aspect at work here; Services and amenities catering to people without cars will only appear proportionately to demand, while people will only choose to downsize or give up their cars as these become available. This already occurs along older established subway routes. It remains to be seen, though, if automated cars (either privately owned or taxis) will be substantially different from what would be populating the road otherwise. I hate to say it, but I can imagine a Lincoln Navigator pulling up to take you 8 blocks at 20 mph.

As for large scale transportation like trains, they are indefensible from terrorists or nutjobs, and the advantages of such scale are diminishing or perhaps completely gone already for most ground transportation, especially if one considers the inefficiencies of aggregating and dispersing passengers from large stations. Perhaps very high-speed rail is an exception as air resistance becomes so extreme at those speeds... Note to readers... 300 mph is an exaggeration...

qt said...

Owwwww. Hope *all* that works out.

Yeah, I can see robot limos. In fact, I expect it. So many better things to do than drive, and you need *room* to do them, right? And that's before you get to the bragging-rights factor...

Scale makes some sense on long trips. Stretching room, if nothing else. But I'd prefer room to sheer speed--especially within arms reach of the ground. And bringing my own car would be a big bonus. Especially if I'm *not* going to an urban destination.

We might be headed for city people with an RV as their only car. Rentals to cover the PRT gaps, so your own "car" is strictly for long trips.

Amusing thought, anyway. Not a lot to do with PRT design, though...

Harald Hygen said...

Hi Dan, I've been following your blog for quite a while from Norway. Thanks for all the good posts!

Form a user perspective one of the biggest advantage of the automobile is that you get in your own private space at home, and out at your destination. The connection between the (self driving) cars / taxis and the PRT system you describe here has a disadvantage. As a commuter my self, I know that changing transportation one or more times is a source of stress, and you can't get deep into Dan's blog while walking from the subway to the bus..

It's possible you've discussed this before I started following you, but what about this:
Have the PRT pods fitting into a "car" module that can run on batteries or gasoline. If you can develop a standardized module the connection between the PRT system and car module can be automatic and seamless. You can drive the car manually until the self driving systems is up to date. The detachment of the car module when entering the PRT system will prevent any extra cost of the infrastructure and energy use. The car modules can be different depending on its use, from golf cart size for inner city to Tesla-style on highway use. An important point here is that the space, the seats, the smell and coffee stains on the table is your own (if you buy a pod)..

All the best!
Harald

qt said...

We have indeed discussed this idea, Harald. In fact, it's what we were discussing here--"dual-mode PRT" is PRT with detachable,independently mobile pods. Most advocates cite it as a solution to the "last-mile problem." I was saying here that the self-driving mini-taxi would greatly reduce the strength of that argument.

Your point, on the other hand, is still a valid one--one I have also mentioned in past comments. The way I phrased it was "we only sleep in our houses--we *live* in our cars." And you're right--Google-mobiles won't solve that problem. Either we change the way we live, or we find a viable solution to this.

As well as what I've called the "honey-do list" problem--that PRT is as bad as taxis for any trip that involves errands and multiple stops.

I've brought these points up, and was amused by the way many PRT advocates would casually say "Oh, they'll just have to adjust." As if forcing the people you're trying to sell this to, to rewrite their way of life to suit *your* vision was no big deal. And then they spent long paragraphs wondering why they didn't have more support...

And now I've gone and forgotten my own points, and you've made them again--including to me! Thank you!

I suspect Dan will be glad to have your insights here as well. He's been hoping for other engineering types to help him think aboit the tech problems, but I've seen that he also values us,too--who think like customers.

Harald Hygen said...

Thanks for the insights QT, I'm currently working on a presentation of PRT to a political party. In addition to Dan's blog I'm following SkyTran. Hope you don't mind this question on your blog Dan? What other promising PRT projects is worth a look?

Dan said...

Dan the Blogger has found internet, for now…
Hello, Gentlemen! Sorry for my absence. I am now back in New Hampshire and “off-the-grid,” except for the library. Smart phones don’t even really work here, so I am mostly concentrating on refining the latest iteration of “Mama Bear” and the associated track, which I do before bed and over morning coffee. It’s a long process!

With the case for automation gaining ground, our battle has shifted to convincing someone amongst the big players that such an effort will have much more upside by taking the plunge and creating an infrastructure that is optimized for such automated people moving. To this end, I believe I can be of most help by wrapping up the SMART platform and summarizing it in a fashion that can be more concisely presented, instead of requiring readers to dig up nuggets from dozens of posts just to get up to speed.

Harald, as you point out, if I understand you, the idea of dual mode takes on new possibilities if part of the vehicle can go back home without you, as opposed to transporting the entire vehicle, which would mean not having the robustness (weight) of today’s cars. Your idea further points to the need for a track that is truly universal to allow autonomous control schemes and ownership models that used to be impractical but may be coming of age. Your anxiety from changing transit modes is, of course, derived from not having a transit system that is ubiquitous. A system that makes enough sense to compete with roads for funding and can provide much more “bang for the buck” would be a system that could grow rapidly into a network with a high likelihood of going the full way from origin to destination for more people more of the time.

This is one of my gripes with Skytran. It involves a very complex magnetic interaction between track and vehicle, so you have to basically take or leave whatever limitations may be inherent in their design, which they never really spell out. Is this really a good system for short haul traffic? What are the feeder lengths and recommended line speeds, turning radii and so forth? It seems like a very good system for longer, high-speed commutes, if the track really is as cheap as they claim, but can it grow into a complex, multi-nodal system? Especially with them being the sole supplier?

One company that is advocating a design framework and philosophy very similar to SMART is Beamways, (Sweden) although they are not open source and have not tested any hardware to my knowledge.
QT, thanks for standing in for me!

qt said...

Ow. I didn't see that side of Harald's proposal.

One of my problems with "pod and sled" dualmode was always the ownership model--you were stuck with whatever "sled" happened to be available at your station, no matter its condition or suitability to what you were doing.

But if you own a sled and send it home to wait for you after dropping you off, you've eliminated half that problem. And some kind of "club" arrangement, or benefits package, or a combination, could reduce the other half. Interesting.

Mark Lee said...

Hey, Your post is very informative and helpful for us. In fact i am looking this type of article from some days. Thanks a lot to share this informative article.

modern office interior design

lee said...

Dan,
I think that the subject of self-driving autos belongs somewhere else. Perhaps I am incorrect but my impression of one of the main goals of PRT is to reduce the number of autos on the streets. Dual mode and self-driving autos do not do that. The development of automated control systems will be a boon to the developers of PRT systems but the autos are in a different arena. If they are developed and accepted has no bearing on the development of PRT. PRT would remove autos from city streets just as buses, trollies, light rail and commuter trains attempt to do. Getting autos into town faster and safer is laudable but the auto is still there and clogging the roads, creating pollution etc. A fully functioning PRT system would remove the auto while making travel safer and faster.
Lee

qt said...

lee,

Actually, the Google-mobile phenomenon is relevant to PRT in at least two ways, discussed here and in earlier posts:

First, the technology developed for self-driving cars has completely changed the PRT paradigm. Originally, PRT was basically an incredibly complex "electric train set" set up, with a central control system micromanaging a huge number of teeny tiny "trains." Dan started out being a bit controversial because he advocated giving the pods more autonomy than was commonly proposed.

But the control and navigation tech developed for self-driving cars means a PRT system can be almost completely decentralized. The pods can choose their own routes, negotiate intersection and switch use, and in general look out for themselves--using a much simplified version of the hardware and software used by Google-mobiles and the like. All you *really* have to do is put up and maintain the track, and add pods as needed. It makes PRT a much more incremental, and therefore viable, alternative.

Secondly, as we are discussing in this post and thread, a simplified autodriven car can provide a *viable* solution to the "last mile problem." If I can call a golf cart to the house and ride it to the nearest station--or buy a pod and an autodriven sled to take it to the nearest station--the system becomes much more accessible, and therefore more attractive to potential riders.

In that case, the autodriven cars could mostly be relegated to local streets and secondary roads, making short trips to the nearest PRT station at lower (and therefore safer and more energy-efficient) speeds. And the PRT system would still be reducing the overall traffic on the city's road system.

Discussing the impact those applications of the technology (and others--I don't doubt Dan is considering some) could have on an overall PRT system seems quite reasonable for a blog like this.

Dan said...

Dan The Blogger responds -

Really well put, QT! I have nothing to add, other than to underline the fact that if auto-driven cars, rented or owned, become a preferred way to get to the nearest PRT stop, this will influence PRT station design and placement, whether we like it or not, and this, in turn, trickles down into the design and economics of the entire system including the vehicles themselves. Like it or not, PRT must be designed for the world in which it is likely to be implemented, not the one we have now.

It is extremely easy to design a system that could be implemented within some kind of benevolent dictatorship, funded by compliant taxpayers, built on land seized by eminent domain, expanded in a logical, multi-year master plan, and able to be operated at a loss until it gets on its feet. The SMART system is designed with the assumption the extremely tough sledding in a free-market driven future, (ie the real world, a couple of years out) and some aspects of it may make more sense when seen in this light, rather than simple mechanical logic.