Sunday, July 7, 2013

Our future with robocars: our future with guaranteed sprawl?

Planners and technophiles have begun to think a lot about autonomous vehicles or “robocars”. Some are even creatively deliberating the possible upsides to this technology for green urbanism and land use. (There are some upsides, supposedly.  See here.)  Most of these visions even entertain a vision of a future where no one owns a vehicle anymore or even knows how to drive one manually.  So great things in store for vehicle share, apparently.  Folks will simply hop on “automated taxis” (ATs) and zip around town purposefully to meetings on dew-drop shaped permutations of MIT podcars. Their attentions no longer demanded at the wheel, the imaginary occupants of these curiously slender cars (even traveling two or three to a lane we hear!) will now spend their time caressing their electronic devices, if not attending teleconferences in their capsule, at least writing a good bit of code, blogging about new products, or dreaming up their next start-ups. The fact that “productivity” benefits for robocars crop up while paying scant heed to their cultural underpinnings and land use implications is, well, interesting.

Whatever the inevitability of automated driving might be, the belief that autonomous vehicles will lead to a future of pervasive carshare is not all that convincing to me.  Yes, I am aware that the percentage of kiddos these days who are eschewing vehicle ownership apparently is on a steady trajectory to reach Edwardian era levels. Should the driving boom be over indeed, however, it is too early to confidently predict the widespread disappearance of vehicle ownership just because people stop “driving”. Nor will robocars helpfully mobilize our efforts to convert infrastructure and development pattern to a car2go utopia, even if they are used predominantly as ATs. I suspect, in fact, the opposite. While carsharing in itself can remove vehicles from the road in per capita terms, we need to better factor here what automated driving represents in two aspects: in the cultural one, especially in how people and machine transition together toward full automation (if indeed they ever manage to), and, secondly, in consideration of the potential commuting dynamics of automated driving.

First the latter aspect.  Possibly the most distracting transport models to talk about in discussing robocars are carshare systems and personal rapid transit (PRT). Distracting because both carsharing and PRT favor compact, walkable urbanism to support them and (at least in their early versions) outright confine their use to certain home areas. They have thus the problem of range. That is exactly the aspect which makes them favor urban and walkable environments. You need to walk to access them. But this is not a geographic limitation facing automated driving at all! In fact, it is a limitation automated driving, or a hybrid version of it, is perfectly suited to overcome. A more strongly correlating transport model to compare with are the mixed-mode carriage systems once proposed in the 60s.  These transport models, now seemingly forgotten, were proposed in the heyday of the PRT-visioning years to serve low density development patterns.  Basically these are like dual-mode PRT systems but adding the de-linkability of the vehicle so that the vehicles can be the conventional gas-guzzling, owner-operated kind.

At their heart, mixed-mode carriage systems were intended to resolve the “modal dilemma”, the impossibility of having an efficient mass transit system that can service far-flung suburbs extensively, flexibly, and cheaply. And so the transit visionaries of the 60s and 70s proposed to create carriage guideways to link cars to belts or put them on trains, thereby lending automobile use some of the virtues of mass transit by escorting linked vehicles rapidly through densely traveled areas. But, as Kevin Lynch pointed out, “this not only requires a very expensive carriage and control on the main routes, but also that individual vehicles be made compatible with that device” (in K. Lynch, Good City Form; Camridge, MA: MIT Press, 1981, p. 423).

One can see the enormous physical barriers to overcome in implementing such a carriage system, and easily deduce good reasons why automobile manufacturers would be averse to adopt it to begin with. Autonomous driving systems, however, not only can potentially virtualize the entire carriage and control system, they remove the need to standardize vehicles for physical linkage. They also can, crucially, decentralize the control system and make it much less computationally complex.  Robocars have the sensory controls needed to “platoon” them together safely at relatively constant high speeds, all you have to add is the vehicle to vehicle communication system to lend platooned traffic a super-human synchronization that prevents it from backing up. Information, in effect, then replaces the “carriage”.  The carriage problem thus is 100% solvable. What’s more, the decentralized aspect of the system means that it can also be modular - adjustable locally so that if one link slows or breaks down (as vehicles are wont to do), the system can more easily handle mishaps safely and insulate a local disturbance from the performance of the whole.

Because the robocar “carriage” is virtual, furthermore, it has an inbuilt flexibility unavailable to any physical system. Thus, for example, access to a high-speed guideway is a much more fluid and safe dual mode transitioning process than its physical counterpart ever could be. So long as spacing is not too tight, vehicles can simply merge into and exit the guideway at any point.  Platooned vehicles merely need to preserve a local “springiness” to receive and dispel individual vehicles in their group so that they will not hurt the overall speed of traffic traveling in the system. If and when local capacity is reached at any point, further merging into the guideway is simply disallowed.  

While it appears complex, the mechanics of carriage-like synchronization become a matter of programming, probably a matter no more complex than can be handled by the programming capacity that is already inherent in automating vehicular driving, I imagine. (I’m no expert of system dynamics and control systems, of course, but reading E.O. Wilson’s descriptions of ant colony communication, I’m quite positive the coding solutions to such complex system-wide control dilemmas will surprise us both in their simplicity and in their elegant results.)

The big hang-up, of course, is convincing humans to yield control of their vehicles, to, well, other vehicles.  It kind of screws with our sense of ownership, control, and even privacy. Yet, I would never underestimate the capacity of early adopters to sacrifice much in their personal sense of self to demonstrate the merit of technological innovations.  Why Google and carmakers are not already testing autonomous freeway platooning is puzzling to me, because it seems to be the low-hanging fruit within the scope of robocar technologies. But, possibly, they may have already realized the immense cultural jump it requires in terms of user adoption. They may have deduced that you need full automation at the outset before people are eventually acculturated to use it on highways. Besides, a dual-mode stage of operation has unique dangers. Google may just not be wanting to waste its time.

Which brings us to the cultural questions surrounding automated vehicle use. Cultural adoption with the robocar guideways described above is easy to imagine incrementally.  If reserved lanes for robocars can be deployed cautiously for early adopters in limited areas as a way to address the most hopelessly congested urban freeways, perhaps by using existing HOV lanes, progressive success with these will no doubt encourage further reservation of travel lanes for the system.  The matter may be a deceptive fix, however, because all it does is displace the congestion to the entry and exit points of the system where guideway traffic will inevitably have to return to surface streets. It will thus drive the need to take even more space for ramps and merging lanes and even creating larger interchange loops to make them robocars keep their speeds going.  In other words, we will be doing (and even adding to!) this kind of scary stuff...

Maybe we can, like, spin robocars right into the streets? A "turbine" interchange improvement project. Source.

I hope I’m wrong.  I do hope robocar control systems somehow prove smart enough to reduce congestion while not adding lanes and speed-fixated freeway infrastructure.  But if so, what’s to stop sprawl at the ends if increased capacity induces demand, as the iron law of freeway capacity improvement dictates? The need to continue improving and adding to limited-access roadway capacity within the city center will certainly not go away once it is clear that the mitigation of highway congestion using the superhuman and decentralized control systems of robocars is feasible. Horrifying as the thought is to me, cities whose current planning trajectory is to improve the speed and capacity of downtown freeways, such as my hometown of Charlotte, N.C., will be vindicated for poo-pooing the freeway-removal, “Ringstrasse” dreams of urbanists such as myself.

The capacity of robocars to extend the geographic range of work commuting and serve the expansion of sprawl should give every land conservationist pause (by the way, land/water resource and wildlife conservation is a primary reason I am an "urbanist").  Sprawl serves travel patterns that are already no longer confined to a metropolitan catchment area but are, increasingly, operating in a regional one. The interurban pulls of the Megalopolis are very real. As a private sector consultant, I’m not only attuned to the scale of far-flung development, I’m also constantly amazed by how much my own industry of urban development continues to transform - even in its operating bases - from a city-centric one into a regionally dispersed one. One of my projects here in Charlotte, for example, is managed by a developer who lives in the Raleigh area, who manages projects both in Charlotte and in Richmond, VA. The general contractor is based (and wisely so) in Greensboro, N.C., and its management team thinks it is nothing to expect its Triad based subs to mobilize teams to Charlotte.  Two hour daily commutes are nothing to all these folks.  You can forget slender pod vehicles. Most of them are the type of individuals who are served only by beefy looking pick-up trucks, as you might expect. And they are operating regionally, like, right now. Because of LEED, the materials that arrive to the job site often have a net transport life cost that is less than that exerted by the workers banging them together on any given day. What do you suppose automated driving represents for these folks? What do you think it means for other industries, however local their attentions?

The potential of another cultural transformation needed to support robocar use is important to note. It has more to do with how we relate to our vehicles. Driver attention and the communication between machine and driver is actually a subtle technical horizon to overcome in the progressive evolution toward full automation.  Innovation in this area is critical because the more dangerous mode of driving is not the human-operated kind and certainly not the fully automated kind but the hybrid in-between form of it where some capacities are automated and not others. (Humans also need to change the way they relate to smart objects, here-to-fore notoriously untrustworthy. But what if we improve the driver-vehicle communication too much?  Will we, in fact, ever get to fully hands-off automation if that happens? What happens if we find a hybrid middle, or even semi-automated “enhanced driving”, more important to human emotion, especially to our notions of safety and comfort? What infrastructure, for example, will we need to build to handle what happens when a car system “crashes” digitally? Will we be caught in a perpetual adolescent lurch toward full automation?)

I don’t think the auto industry will also ignore the latent capacity of enhanced communication to feed our endless ability to personalize our vehicles.  With enhanced communicative traits, won’t vehicles, in time, become more pet-like? Remember KITT? Enhanced to discern your mood and to chat cheerily with you, they may become rather compellingly ownable extensions of personal and family identity, it seems to me.

Perhaps robocars can be enhanced for share-ability by lending them the communicative qualities of Jane Jacobs’s “public characters” - quirky, talkative but not nosy, and precariously self-maintained.  People may come in time to recognize some of their favorite AT personalities and create public nicknames for them (especially if some units are known to be moody).  Preserving and caring collectively for them then better becomes a kind of public trust. They might even help with our sense of community. But if this be a tactic to get buy-in for ATs, not only is it strange to think it can replace vehicle ownership, we’d still have to ask toward what end if we succeed. The green merits of ATs, while lessening the need to own and park cars, are actually quite dubious compared to car2go enabled urbanism since ATs also remove the need to walk to parkspots. Like I said, that is the barrier automated driving is good at removing. Think of an AT fleet not only as carshare with valet service but carshare for the cul-de-sac.

For all those reasons, I don’t think the green advantages of carshare will lend themselves necessarily to our future with robocars. Neither in curtailing vehicle ownership and storage needs much and especially not in changing our infrastructural promotion of sprawl. In fact, I fear dedicating lanes to high-speed robocar commuting may endanger reserved BRT lanes, due to political pressure to convert them to an automotive use that is perceptually “equivalent” to rapid transit. As Jarrett Walker points out, exclusive lanes of successful BRT systems appear often as “empty” (and thus implicitly inefficient and share-able) to vehicle owners.   

The only thing robocars could possibly guarantee to conserve our environment is allow us to park more efficiently in space-saving ways, but how in the world could that possibly offset the exurban sprawl they will no doubt induce?  Let me know if I’m missing something here.