Friday, September 16, 2016

PRT is Dead. Long live PRT!

With the various advances that have come to transportation lately, PRT has lost its monopoly on many of the virtues that once made it unique. It is, perhaps, natural to wonder if PRT’s time has come and gone. Once upon a time only PRT could manage its traffic because the vehicles were centrally controlled. Now, with traffic apps, roadside electric billboard alerts, etc., drivers by the thousands can increasingly be alerted of upcoming congestion and change their routes accordingly. In the very near future cars will be “talking” to signs, traffic lights, the cloud and to each other. In the past the whole idea of a driverless vehicle was unthinkable without a dedicated track or guideway, simply for safety reasons. That is clearly changing. Once, PRT alone made electric vehicles practical, because an electrified rail eliminated the need for the heavy lead-acid batteries that made electric cars slow and short-range. Another advantage gone. Does all of this mean PRT is dead? Can PRT’s essential advantages can be had in different ways? Let’s dig a little deeper.

One must first realize that progress has not been confined to cars alone. PRT’s centralized control scheme was a liability as well as an asset – A single glitch could strand everybody. Autonomous driving benefits PRT right along with automobiles. This is especially important in that it facilitates variable speeds which, in turn, enable tighter turns and faster straightaways, making PRT much more versatile. Those sweeping, graceful turns once depicted in PRT promo material were far from an asset – who would give up their valuable corner property without a fight? Advances in batteries benefit PRT as well. Now there are many “third rail” options including none at all, being a discontinuous, a charge-on-the-go system, supplementing solar collectors, etc. Wireless connectivity, combined with more “intelligent” PRT control makes for a very robust and fault-tolerant system. In other words, any PRT system that is devised today can be far superior to past designs by almost every metric.

Although many designs still represent a basically workable, even elegant transit alternative, enough has changed in the technological landscape to justify a re-think of many key aspects of the whole proposition. It’s not just self-driving technology and batteries. It’s smart phones, the cloud, how things are prototyped and manufactured. The world itself has changed, including people’s expectations and behavior. About the only thing that hasn’t changed is the need. The fundamental problem with “surface” transportation is that paths inherently cross, leaving no alternative but for someone to stop and wait, whether you’re a pedestrian, train, on bicycle or on horseback. The large size and weight of commercial vehicles on our current roadways curtails extensive use of curative overpasses and even when elevation is absolutely necessary, the high inertia of these vehicles is such that massive cloverleaf designs must be created or they would simply skid off the road at reasonable speeds. This, with the shrinking amount of available ground-level space, ensures troublesome bottlenecks. Self-driving cars, alone, can never solve this dilemma. Private passenger vehicles do not require anything close to such a heavy infrastructure, yet they comprise the lion’s share of vehicular traffic. There is, therefore, great advantage to be gained by giving such smaller vehicles a modern, non-stop infrastructure all their own, which can also eliminate the need for plowing, policing, and save lives. So if a smaller gauge, elevated (and presumably automated) system for moving people is still needed what, specifically, has changed?

PRT has always been a mix of utopian dream and practicality. You walk a short distance to a station, there are private vehicles waiting, and you get whisked away, non-stop, to your destination. But wait: A “short distance?” How short? “Private vehicles waiting?” Every time? Isn’t that kind of wasteful, having that much precious investment sitting idle? What about vehicle capacity? Should all vehicles that will usually carry one be built for four, “just in case?” And what about “non-stop?” If a vehicle can pick up a passenger or two without adding substantially to the travel time, isn’t it kind of inefficient not to? Or what about “To your destination?” How close are we really talking about? Above all, is this practical as a business?

Unfortunately, the whole notion of PRT becomes very iffy when many important stations have yet to be added, because it is that much less likely that there will be potential passengers, at any given time, within walking distance of a given station. Trying to start a PRT system without a multitude of destinations is like opening a grocery store that only stocks pet food, diapers and milk and expecting it to grow from there. No selection, no customers. I suspect we all agree, however, that PRT would prosper fabulously were it extensively built-out to the point of having stations on every corner. That would be a “Walmart” of destination choices! But how do we realistically get there in stages? This has always been the fly in the ointment and is where the latest technological trends may just come to the rescue, even if the remedies aren’t exactly what a PRT purist would like. Let’s start with the first part of the supposition above, walking that “short distance” to the station.

It is no secret that Uber is working on self-driving cars. Yet as the recent fatal crash of a Tesla on autopilot points out, there will be a lot of bumps along the way, especially in terms of liability and ethics when it comes to driverless cars going lethal speeds.  For robotaxis, it is more likely that completely driverless vehicles will cut their teeth on the slower, short-haul market for quite some time before attempting to navigate amongst drivers who themselves are often driving unsafely, sometimes on foggy, icy or flooded streets. A preferable start for an Uber would be, therefore, somewhere that would leverage the convenience of a ride hailing app with a greater value than a short, slower ride would ordinarily suggest. Such situations already exist in cities with long established and heavily used transit systems, where some neighborhoods are just out of comfortable walking distance to the nearest station. Assigning a few robotaxis to each station would extend a transit system’s reach. Better yet, however, would be to service a PRT station, because that would mean that an entire trip could be arranged in a single step. In other words, a robotaxi/PRT/robotaxi trip could be arranged by phone, giving the full door-to-door service of a traditional taxi but with the non-stop, flyover capability of PRT, hopefully for the lion’s share of the trip. This would require far fewer PRT stations than the traditional grid approach, with the PRT simply acting as a wormhole from one part of the city to another. Such a partnership is as good for Uber as it is for PRT.

Then there is the matter of sharing the ride via Group Rapid Transit, or GRT.  Automated self-driving technologies enable a mix of vehicles. There can be 2 seaters, 4 or 6-10 seaters or cargo vehicles all sharing the same track. A cloud-based phone app could do more than simply synchronize the ground transportation for a PRT passenger. It could also pair passengers with vehicles “on the fly” to create the most efficient shared routing. Since nascent networks will necessarily have only a few major destinations to start, it is vital to implement a business model that can survive this stage. Such stripped-down networks tend to lend themselves more to GRT than PRT, since it makes little sense to send large numbers of people to the same destination in separate vehicles. Readers of this blog will note that the concept of a multiple-tier, yet compatible track system (Baby, Mama and Papa bear track) has been proposed herein, and GRT track is of the “Papa Bear” variety, which supports high speeds and heavier weights, but not steep slopes or tight turns that would be best for confined, densely developed areas. Although not as flexible as a smaller track, it is assumed that many important destinations are accessible by major highways anyway, and such track would generally follow these or other uncontested routes. In a more mature system these would form an arterial backbone for smaller, yet compatible PRT track and station designs that are sized to more comprehensively serve neighborhoods where space and flexibility is at a premium.

This is where cloud-based “intelligent” routing and scheduling really makes a difference. Consider the compromise reflected in all forms of shared, scheduled transit: Running infrequently makes for full vehicles but discourages use of the system while running frequently promotes system use but it may operate at a loss because the vehicles are nearly empty. With app/cloud-based systems, however, the need can be monitored in real time and responded to accordingly. This strongly implies an approach that incorporates multiple vehicle sizes, something that would make sense anyway, from airplanes to buses, if only such vehicles could be staged “at the ready,” and there was a means to accurately predict occupancy. A paradigm of distributed, algorithmically anticipated, haled (rather than scheduled) vehicles makes this possible for the first time, essentially turning everything we thought we knew about mass transit on its head.

In a fully automated, cloud-based deployment scheme, artificial intelligence would be used in real time to choose both vehicle type and destination to best eliminate waiting and minimize travel time while maximizing throughput. Unlike a city bus, which often has a sign that shows route/destination information, A GRT vehicle’s sign (and destination) might change as it pulls into a station based on the more democratic principle of majority rule.  Perhaps such a system could even arrange a passenger transfer to another vehicle, maybe even calling that passenger by name and giving instructions. For that matter the vehicle would also “know” if a passenger is making a mistake and getting off at the wrong station. The system could even respond to passenger preferences and histories, such as not pairing an unaccompanied woman or minor alone with a stranger, or only pairing students of a local college with other students. Hmm… This might give new meaning to the term “speed dating”! Seriously though, bad experiences can be reported within a passenger app, and with cameras becoming ever cheaper and clearer, this all bodes well for safety and passenger satisfaction even without the watchful eye of a driver.  

All of the above stands in stark contrast to the “everything or nothing” proposition of traditional PRT, and we all know which of those two cases has come true to date. Is PRT “dead?” Perhaps, in a way. If this is so than it is death by a thousand cuts… a business model that is not quite compelling because dozens of very minor (though often solvable) factors add up to a no-go, especially in the early stages when there are too few stations to create the “network effect.”  Still, besides creating a dizzying array of choices, our current technological prowess has made the path towards those original goals more likely and achievable, not less. At present it seems that GRT (rather than PRT) is probably the way to start because the drawbacks of group travel have been largely eliminated and GRT offers a better business model for the fewer stations that are an essential stage in building out a network. GRT, like PRT, with a robotaxi at each end, offers true door-to-door service, ideally with near zero wait time. It is also notable that even with a GRT system there would be naturally be a need for smaller vehicles. Late at night, for example, it would make little sense to use vehicles that are sized for rush hour. This bodes well for the prospect of arterial track sections that are eventually populated by both PRT and GRT vehicles. 

Any ground-based leg of the trip, of course, will still experience traffic delays. That, for PRT (and elevated transport in general) is a good thing, as it will encourage expanding the aerial network into more and more areas, and encourage players like Uber, Google, or even Ford to enter the elevated transit business. The contrast between crawling along on the ground and flying over the city will be like night and day, and it is hard to imagine that they will not see the light. It is also noteworthy that future robotaxi logs would give solid statistical evidence as to where the elevated track should expand to next, by predicting how such a “spur” would perform monetarily, removing the financial risk of such an expansion.

PRT does seem dead or dying as a singular, all-inclusive product/service, in that the old model is  analogous to treating cars and roads as a single thing, when they are many: There are many types of roads and many types of vehicles that drive on those roads and that’s a Darwinian strength. The specific vehicle/track/operational methodologies of many past PRT designs were often tightly integrated to compensate for technological limitations of the day and so now require total revision in the face of the game changing advent of autonomous vehicular control. In this new world the track itself is the most important part to get right, because it is, ultimately, the limiting factor. Track that is too expensive, uses too much real estate, requires stations that are too expensive, is noisy, is affected by weather, cannot easily be modified for various loads or spans, cannot support high speeds, cannot be quickly and inexpensively  constructed or modified, etc., won’t fulfill its promise, or the promise of anything running on it. Above all it needs to be flexible, and that means being designed for much more than PRT. It seems increasingly likely that vehicles from multiple venders may be the business model that is most likely to flourish in the ecosystem of grade separated, automated transport, so we have to keep the track simple.

The roll of the “control system” has clearly changed. There has always been some degree of autonomy in PRT, say to slam on the brakes to avert a collision, for example. With newfound autonomous capabilities however, centralized “control” becomes, more and more, merely a role of scheduling and traffic optimization. Under such a scenario, the central computer basically makes suggestions in real-time that are normally followed, so it is essentially the same as central control. The difference is that the vehicles are quite capable of working independently, just at a more cautious pace. This may seem like splitting hairs, but it is an important distinction from a business perspective, because it redefines what role a PRT company would have in continuing operations by reducing an important role to a discreet, probably largely cloud based, software layer.
  
What if, we might dream, a city were to say to a Google, Tesla, Amazon or an Uber, “If we provide this track and these stations will you design and build robotic vehicles for it? Or perhaps provide the central traffic optimization (control) layer? While it might seem, at first glance, that such companies would have little to gain, it is actually a pretty interesting sandbox, as described in the previous post, and the cost of prototyping and mass producing the actual machinery is falling every day. There has always been a natural dichotomy in PRT between that which is naturally a city’s business and what a PRT company can be expected to produce and perform. Autonomous vehicles enable a “dumb” track that requires less oversight, is less likely to become obsolete, is less tied to a specific vehicle manufacturer, and is therefore generally less burdensome and risky to a city. This, in turn, relieves vehicle manufacturers from a business they want no part of, that being planners and builders of urban infrastructure. The PRT provider really becomes more of a go-between, a builder of partnerships. Everybody’s happy!


The PRT of old seems pretty much dead... if it was ever really alive to start with. Loosely defined, however, where we are simply referring to smaller, haled robotic vehicles operating on a network of lightweight, elevated tracks... well that, my friends, is alive and well. Long live PRT!