Robocar Cheerleader Disses Transit PDF Print E-mail
Written by Administrator   
Wednesday, 25 January 2012 10:47
By Michael D. Setty

At the "How We Drive" blog (http://www.howwedrive.com/2012/01/24/systemempathy-in-transit/#comments) leading robocar cheerleader Brad Templeton (http://www.templetons.com/brad/robocars/) dismisses the need to improve transit:

Tom, now that you’ve seen the speed at which robocars are developing, you may want to consider one of the possibilities I have been investigating, which is the decline of public transit. While there’s no assurance that the world will switch to them, robocars that can self-deliver makes it much more marketable to have people ride in lightweight, 1 and 2 person short range electric city vehicles. Today you can’t sell them but they are perfect as cell-phone-summoned robotic taxis...

...It’s my view that almost all the rules of transit will be erased and rewritten in the next few decades, but transit and urban planners don’t even have it on their radar.


I understand completely when anti-transit activists such as Randal O'Toole or Wendell Cox dismiss transit for conservative/libertarian "philosophical" (sic) or "cultural panic" reasons (for an excellent discussion of the latter phenomenon, see http://www.cnn.com/2011/OPINION/04/21/harrod.high.speed.rail.trains/index.html), but Templeton's argument that technology will solve what are essentially human social and economic problems is the weakest I've seen yet.

My response to Templeton's comments at the How We Drive blog are "below the fold."



As perhaps the leading cheerleader for "robocars," it is reasonable that Brad Templeton quickly dismisses the alternatives, such as dramatic improvements to public transit in the U.S. However, his arguments for neglecting transit improvement are weak, at best.

As a transit supporter and robocar skeptic, my objections are listed below, from the least to most significant.

First, assuming robocar technology is as robust as Brad claims it is, you then run into the "quality of components" problem. Robocar proponents make huge claims for the idea, which they must admit will place public expectations for safety improvements over existing automobiles at least an order of magnitude, if not 50-100 times higher.

But actually obtaining such results will depend on extreme reliability of the hardware, particularly the computer hardware and associated complex mix of vehicle sensors. In turn, such levels of reliability will require very high quality components, e.g., "military" or "Mercedes" grade as opposed to "Hyundai" grade. Mass produced sensors may cost, say $50 to $100 each, each vehicle requiring a dozen or more. But the most reliable components typically cost an order of magnitude more. In this regard the problem resembles the current state of the hard drive market; mass market 1TB drives retail around $100-$150 but much more robust "mission critical" but much lower capacity mechanisms cost several times more, on average.

I'm certain Brad will argue that quality and reliability will improve over time, but the current state of computer and hard drive technology has taken 3+ decades to evolve to its current state. A similar timeframe is likely with complex technologies such as robocars. And I'm not even commenting about the state of software! After nearly 30 years of development, Windoze is still a big kludge! LIke the original adoption of the automobile, robocars will require at least 2-3 decades for full adoption, probably more if you consider that "full adoption" in the U.S. of the automobile didn't fully occur until the 1970's.

Second, I'm still not convinced the state of artificial intelligence (AI) is nearly as robust as Brad has claimed elsewhere.

At a minimum, several years of "real world testing" will be required to prove such claims in a wide variety of complex urban environments--assuming any cities out there willing to be guinea pigs. In Nevada driving across the desert is one thing; the Las Vegas Strip or South Virginia Street in Reno is another.

Third, there is the "human" problem.

After nearly two decades of effort, the government has managed to convince 90%+ of people (well, at least in California) to "buckle up." One of the standard pitches for robocars is that one won't have to pay attention to driving, so one can undertake other activities during the ride. However, in the same manner that having anti-lock brakes cause many to drive "closer to the edge," overconfidence in the abilities of robocars could cause many to forsake "buckling up." With such a "human" problem, the safe braking distance margin would have to be greatly increased compared to EXISTING automobiles, if people aren't to be splattered on the dashboard. Instead of a dramatic increase in road capacity, this would greatly reduce capacity.

The alternative would be robocar safety interlocks that refuse to move unless seat belts are attached--also requiring seat weight sensors. In turn, how does the computer determine whether that weight in the seat is a human, a dog, a package, groceries, or a baby/small child (and how does the robocar enforce child car seats?) And so on. Remember the 1970's seatbelt interlock fiasco that the public hated with a red hot passion?

Fourth, the robocar seatbelt problem raises the issue of how the government would regulate robocars. Currently, driving is perhaps the most highly government-regulated activity the average person undertakes on a regular basis. Given the complexity of robocars and the extremely high level of safety expectations, government regulation of the technology is likely to be at a much higher level. Modifying vehicle systems, e.g., "tampering" in bureaucratese, could conceivably become a felony. Insurance companies would certainly want a complete record of everything when investigating robocar accidents and failures, as certainly would the government.

Cellphones with GPS certainly raise an entire host of privacy concerns which could pale in comparison with those raised by robocars. Will the interior cameras robocars use to determine whether that weight in the seat is a human, a dog, a package, groceries, or a baby/small child also be another revolutionary tool for cops and divorce lawyers?

Fifth, why not simpler technical solutions such as still-human driven "city cars" resembling beefed up golf carts? Unlike robocars, this idea has been prototyped for years in Palm Desert, California and a few other communities. Why not simply require new developments--particularly employment centers--to be pedestrian and bike friendly, making them transit friendly as a bonus? Not every problem has a "high tech" solution or long-proven technology that solves the problem already exists.

Finally, before society makes the leap towards wide-spread adoption of robocars, why can't we slow way down and consider if such technology will create the sort of cities, towns and suburbs we want. Or will we allow robocars to totally remake the landscape to meet the peculiar needs of robocars, in the same way we did to accommodate the human-driven automobile? As someone who prefers to put pedestrians, e.g., PEOPLE, first, I consider that there may be a suitable niche for robocars in our transportation system, but not nearly as widespread as in the wet dreams of many techno-fetishists.
Last Updated on Wednesday, 25 January 2012 11:01