By Lambert Strether of Corrente
Spoiler alert: Pretty badly. Let’s start with the hypester’s hypester, Elon Musk:
Musk starts cranking out TED Talk-ready sound bites. “. We know exactly what to do, and we’ll be there in a few years. We’ll take autonomous cars for granted in quite a short period of time,” he says matter-of-factly. “You’ll be able to tell your car, ‘Take me home,’ go here, go there, or anything, and it’ll just do it. It’ll be an order of magnitude safer than a person. In the distant future, people may outlaw driving cars because it’s too dangerous. You can’t have a person driving a two-ton death machine.”
John Zimmer, co-founder of Lyft:
Autonomous vehicle fleets will quickly become widespread and will account for the majority of Lyft rides within 5 years. , private car ownership will all-but end in major U.S. cities.
Travis Kalanick, Uber, interviewed by Biz Carson:
[KALANICK:] I think it starts with understanding that the world is going to go self-driving and autonomous. Because, well, a million fewer people are going to die a year. Traffic in all cities will be gone. Significantly reduced pollution and trillions of hours will be given back to people — quality of life goes way up. Once you go, ‘All right, there’s a lot of upsides there’ and you have folks like the folks in Mountain View, [California,] a few different companies working hard on this problem, this thing is going to happen.
[CARSON]: How soon will self-driving cars realistically be a significant portion of Uber’s fleet?
[KALANICK:] That is the trillion-dollar question, and I wish I had an answer for you on that one, but I don’t. Right? I have to make sure that I’m ready when it’s ready or that I’m making it ready. So, I have to be tied for first at the least.
So, three Silicon Valley innovators talking their books. And Ford:
Ford CEO Mark Fields said two weeks ago it will have a fleet of completely autonomous taxis operating in an unnamed city .
Although GM is slightly more sensible:
[Dan Ammann, GM’s president] sees self-driving cars becoming a reality as “a series of developments, opening up to broader to broader as we go.” He would not name a specific timeline. (“There’s no single point answer to that question,” he said.)
(We’ll see the framework that Ammann’s thinking fits into in a moment.)
Let’s pause for a moment to get the taste of hype out of our mouths. In the last post on self-driving cars (autonomous vehicles), I asked whether they could even be marketed, and thought not; people want others to have cars whose ethical algorithms might sacrifice passengers for the greater good, but want cars that protect passengers at all costs for themselves. In this post, I’m trying to get a reading on whether the tech is really there. In future posts, I’ll look at government regulations, what the business models for selling self-driving cars might be (if indeed they are to be sold, as opposed to being rented), effects on political economy (income inequality, public works, insurance, jobs), incremental approaches (trucks on highways first), and social benefits (for example, lives saved).
Back to the tech. Everybody loves a taxonomy, and as it turns out, there are two of them for self-driving cars: the National Highway Traffic Safety Administration’s (NHTSA), and the Society of Automotive Engineers‘. The NHTSA’s levels are 0-4 (total five) and the SAE’s are 0-5 (total 6). Pleasingly, therefore, if you hear “level 4,” you don’t automatically know which standard is meant. But if you hear “level 5” you know that the SAE is meant (even though the NHTSA has five levels). In what follows, I’m going to use the SAE’s, because the SAE, even if not an international standards body, is at least North American. Here’s a handy chart of the levels (the full size version is here):
And here’s a technical person’s explanation of the chart, from Dr. Steve Shladover of Berkeley’s Partners for Advanced Transportation Technology. I’ve added the SAE level names in bold:
[Driver Assistance]: So systems at Level 1 are widely available on cars today.
[Partial Automation]: Level 2 systems are available on a few high-end cars; they’ll do automatic steering in the lane on a well-marked limited access highway and they’ll do car following. So they’ll follow the speed of the car in front of them. The driver still has to be monitoring the environment for any hazards or for any failures of the system and be prepared to take over immediately.
[Conditional Automation]: Level 3 is now where the technology builds in a few seconds of margin so that the driver may not have to take over immediately but maybe within, say, 5 seconds after a failure has occurred.
So in theory you could be reading a book or you could be playing a video game or surfing the web on your tablet or doing something else like that while driving and then the system has a problem and it beeps you and it says “You need to take over” and you have to now turn your attention back to driving.
That level is somewhat controversial in the industry because there’s real doubt about whether it’s practical for a driver to shift their attention from the other thing that they’re doing to come back to the driving task under what’s potentially an emergency condition.
[High Automation]: Level 4 gets more interesting because this now where the system has enough internal redundancy that it can take over from itself when it has a fault.
So it has multiple layers of capability, and it could allow the driver to, for example, fall asleep while driving on the highway for a long distance trip.
So you’re going up and down I-5 from one end of a state to the other, you could potentially catch up on your sleep as long as you’re still on I-5.
But if you’re going to get off I-5 then you would have to get re-engaged as you get towards your destination.
That could also be a low-speed shuttle that would operate within a confined area, like a retirement community or a resort or shopping complex, where the interactions with other vehicles might be limited so that that helps keep it safe.
[Full Automation]: Level 5 is where you get to the automated taxi that can pick you up from any origin or take you to any destination or they could reposition a shared vehicle. If you’re in a car sharing mode of operation, you want to reposition a vehicle to where somebody needs it. That needs Level 5.
Note that Musk, Zimmer, and Kalanick all treat Level 5 (Full Automation) as a done deal. That’s what they’re selling, Musk “in a few years,” Zimmer “within 5 years,” Kalanick “What I know is that I can’t be wrong,” and the Ford CEO Mark Fields “by 2021.” Shladover disagrees:
Contrary to Musk and many of the most prominent advocates of autonomous cars, Shladover insists that so-called Level 5 vehicles—robocars that require no human input—are not on the horizon. “. And I say the same thing to students,” he says. “Merely dealing with lighting conditions, weather conditions, and traffic conditions is immensely complicated. The software requirements are extremely daunting. Nobody even has the ability to verify and validate the software. I estimate that the challenge of fully automated cars is 10 orders of magnitude more complicated than [fully automated] commercial aviation.”
Herman Herman, director of the Carnegie-Mellon University Robotics Institute, disagrees as well:
With autonomous cars, you see these videos from Google and Uber showing a car driving around, but people have not taken it past 80 percent. It’s one of those problems where it’s easy to get to the first 80 percent, but it’s incredibly difficult to solve the last 20 percent. If you have a good GPS, nicely marked roads like in California, and nice weather without snow or rain, it’s actually not that hard. But guess what? To solve the real problem, for you or me to buy a car that can drive autonomously from point A to point B—it’s . There are fundamental problems that need to be solved.
Finally, I advise readers to read the testimony of Dr. Mary Cummings before the Senate Commerce Committee on March 15, 2016 in full (like Shladover and Herman, Cummings has devoted a large portion of her professional life to tacking significant problems in robotics). Here’s Cummings on some of the technical issue:
While I enthusiastically support the research, development, and testing of self-driving cars, as human limitations and the propensity for distraction are real threats on the road, I am decidedly less optimistic about what I perceive to be that are absolutely not ready for widespread deployment, and certainly not ready for humans to be completely taken out of the driver’s seat.
Here are a few scenarios that highlight limitations of current self-driving car technologies: The first is operation in bad weather including standing water on roadways, drizzling rain, sudden downpours , and snow. These limitations will be especially problematic when coupled with the inability of self-driving cars to follow a traffic policeman’s gestures.
Another major problem with self-driving cars is their vulnerability to malevolent or even prankster intent. Self-driving car cyberphysical security issues are real, and will have to be addressed before any widespread deployment of this technology occurs. For example, it is relatively easy to spoof the GPS (Global Positioning System) of self-driving vehicles, which involves hacking into their systems and guiding them off course. Without proper security systems in place, it is feasible that people could commandeer self-driving vehicles (both in the air and on the ground) to do their bidding, which could be malicious or simply just for the thrill and sport of it.
And while such hacking represents a worst-case scenario, there are many other potentially disruptive problems to be considered. It is not uncommon in many parts of the country for people to drive with GPS jammers in their trunks to make sure no one knows where they are, which is very disruptive to other nearby cars relying on GPS . Additionally, recent research has shown that a $60 laser device can trick self-driving cars into seeing objects that aren’t there. Moreover, we know that people, including bicyclists, pedestrians and other drivers, could and will attempt to game self-driving cars, in effect trying to elicit or prevent various behaviors in attempts to get ahead of the cars or simply to have fun. Lastly, privacy and control of personal data is also going to be a major point of contention. These cars carry cameras that look both in and outside the car, and will transmit these images and telemetry data in real time, including where you are going and your driving habits. Who has access to this data, whether it is secure, and whether it can be used for other commercial or government purposes has yet to be addressed.
(I really like the way Cumming’s mind works; several nasty twists of thought.) And then there’s the reality of insufficient testing beneath the hype:
In my opinion, the self-driving car community is woefully deficient in its testing and evaluation programs (or at least in the dissemination of their test plans and data), with no leadership that notionally should be provided by NHTSA (National Highway Traffic Safety Administration). Google X has advertised that its cars have driven 2 million miles accident free, and while I applaud this achievement, New York taxi cabs drive two million miles in a day an a half. This 2 million mile assertion is indicative of a larger problem in robotics, especially in self-driving cars and drones, where demonstrations are substituted for rigorous testing.
And the reality of testing results that are not disclosed:
But there are many known knowns in self-driving cars that we are absolutely aware of that are not being addressed or tested (or test results published) in a principled and rigorous manner that would be expected in similar transportation settings. For example, the FAA (Federal Aviation Administration) has clear certification processes for aircraft software, and we would never let commercial aircraft execute automatic landings without verifiable test evidence, approved by the FAA. To this end, any certification of self-driving cars should not be possible until manufacturers provide greater transparency and disclose how they are testing their cars. Moreover, they should make such data publicly available for expert validation.
Verdict: Self-driving cars are not ready. Nowhere near ready, despite what Musk, Zimmer, and Kalanick are saying.
Returning, then, to the question of the question of the NHSTA/SAE levels. Technologist Brad Templeton writes:
Numbered levels strongly suggest an ordering or hierarchy to a technology that almost surely will not evolve in the manner laid out. The levels create an expectation of evolution in this direction and also an expectation that each level is a superset of the one below it. Regulators, press and the public are led to expect this progression by the levels, and may even write rules that demand it.
The levels, in other words, imply a teleology, with inevitable progress to Level 5. But as we’ve seen, it ain’t so. What is far more likely is incremental progress in automotive computing (which is not at all the same as building an autonomous vehicle. From Carnegie-Mellon Engineering Magazine:
As self-driving technologies mature, they are gradually assimilated into auto production. Adaptive cruise control and parking assist are commercially available now. In the 2020s, cars will have traffic jam assistance and virtual valet, a feature that enables cars to park themselves via a smartphone app.
“Features of automated self-driving cars will appear , with vehicles eventually driving themselves. This will make the cars affordable and encourage public adoption,” says Raj Rajkumar, a professor of electrical and computer engineering and the co-director of the GM Collaborative Research Lab at Carnegie Mellon.
And the incremental approach, as we have seen, is the one GM, as opposed to Ford, is taking. One last point. From Mary Cummings and Jason Ryan, TR News May-June 2014:
The majority of the promises and benefits will likely only be realized when all cars are equipped with these advanced technologies, enabling NHTSA’s Level 4 [SEA’s Level 5] of fully autonomous driving.
In other words, Kalanick’s “millions of lives” won’t get saved in our lifetimes, because Level 5 isn’t happening any time soon; Shladover’s Level 4 “low-speed shuttles” sound like a solid little product, but not exactly what Kalanick has in mind, eh? What that implies is that self-driving car technology companies should be valued a lot more like GM, or Delco, rather than at the stratospheric prices of Silicon Valley unicorns.
 The nice thing about standards is that there are so many of them!
What I would like is an explanation of how networked vehicles would necessarily ease urban traffic conditions.
If during high flow periods, there are certain number of people trying to get to a certain number of places at the same time, and they are mostly using single-occupant vehicles, how much difference would it really make if all of the vehicles were driven by a network?
I mean, sure, the spacing could be somewhat tighter, but urban traffic flow during peak hours are already pretty tight. I don’t think the overall vehicle density would be all that much higher.
And sure, the vehicles could be somewhat smaller and more lightly built, but mostly that would just save energy. Okay, parking density could be higher with smaller vehicles, and that could save some street space for more flow.
However, part of the sales pitch for network vehicles is that the vehicles return-to-base and so much less parking would be needed overall. But–that would mean the networked vehicles actually spend more time in the traffic, increasing the overall traffic volume.
So back to the first question: how would network vehicles eliminate urban traffic problems. Unless you’re going multiple-occupancy per vehicle, I just don’t see how it adds up.
We already have a multiple-occupancy model for urban transport: buses and various kinds of off-grade commuter railway.
When I was in Cairo in 2005, what I saw on the streets was basically a traffic flow composed almost entirely of taxis and minibuses. The traffic was like blood circulating constantly through the city. Most of the cars had more than one passenger, and the minbuses were usually crowded. The drivers were expert, alert and enterprising. The flow was constant, spacing between the moving vehicles was as tight as it could be, and little area was wasted on parking. For what it was, the Cairo traffic system was relatively efficient.
Nevertheless, Cairo street traffic is jammed 21 hours out of 24. the Cairo Metro was a much faster and more pleasant way to get around the city, at least in its area of coverage.
What I’m getting at here is that unless you have off-grade mass transit with high occupancy, it doesn’t really matter much whether your individual low-occupancy street vehicles are networked, or human-driven.
I guess for people who both live and work in low-density suburbs, and who commute almost entirely on highway systems, would have at least a less tedious and exasperating commute if their car was networked. They might only need to drive “first mile” to the onramp, and “last mile” to their suburban office park.
But again, when you get traffic jams in a suburban commercial area, it’s simply due to an excess of low-occupancy vehicles, all trying to go to the same places at the same time. I don’t see how it would matter much whether those low-occupancy vehicles are driven by networks or by humans, except in the long run, in terms of reducing parking space allocation in suburban shopping districts. But would those space savings translate into additional road space. On urban streets, no parking means extra traffic lanes. But if a suburban “big box” store deletes its hectares of parking lots, does that translate into additional road space?
Unless somebody can explain to me the real net savings in traffic expenditure, all I’m seeing in this networked vehicles trend is another kind of rent-seeking through forced obsolescence, i.e. prevent people from driving their own vehicles and force them to purchase the networked product.
There is a real benefit possible from closer spacing of networked vehicles. The following diagram of flow vs flow density is illustrative:
Essentially, traffic has two modes: Free flow where increasing traffic density has increased flow of traffic, and congestion where increasing traffic density leads to decreased flow of traffic. Congestion happens because people need to keep some minimum distance from the car in front at high speeds in order to maintain safety, and this leads to a cumulative effect where if the traffic density gets too high then people slow down to maintain safety.
The promise of networked cars is that it would shift the critical density between free flow and congested traffic towards higher traffic densities, because ostensibly automation could allow for less distance between cars. In fact, this is true and we already have this technology vis-a-vis trains. I still maintain that for most of the country (excluding rural areas) better public transportation investment is the right solution for our traffic problems.
*Sigh* This is not the reality of a lot of traffic congestion. See New York City from Thanksgiving to Christmas. I can’t remember the name of the principle from physics, but when you take moving particles in a tube and make the tube smaller, you get backup in the big tube behind the choke point behind the smaller tube. Most traffic congestion is like that: slowdowns due to lane reduction (accidents or construction), needing to go through toll booths, or cars backed up at important exits.
I’m not sure if this is what you are meaning, but Braess’ paradox also puts lie to the notion that a physical network will be necessarily improved by automation.
I think youre trying to recall Venturi Effect from your HS physics class.
I think you are also expressing the sentiment that any idealized assumptions of polite well ordered laminar flow of particles (vehicles) breaks down when applied to trafficjams in NYC –think slow chaotic flow of angry particles giving eachothe the finger before finally ordering and entering the throat at the jam -the three guys watching the bobcat operator digging up the ruptured pipe.
I think inreality, the exogenious variables of life will foil any algo attempt to bring order to traffic in the brownian motion of a clstrfck traffic jam in NYC. To me the notion is laughably absurd, now crush the vehicles bumper to bumpertoeliminateany possibility of maneuvering
Heres a fun one
In our study we look at the driver aggressiveness as a result of lane-drop bottleneck from 3 lanes to 2 lanes. In normal circumstances drivers absolves themselves to the two proceeding lanes far upstream of the bottleneck. This does not cause much traffic jam in case of a lane-drop bottleneck. An aggressive driver on the other hand will take advantage of the free third lane and drive all the way to merge at the lane-drop location. This behavior forces other drivers to emergency braking and as a result traffic jam forms from the lane-drop point and grows upstream of the highway…
AND you get to use the phrase .spatiotemporal behavior
It’s not quite Brownian motion, but it’s not laminar either.
Isn´t there also the possibility for Jevon’s Paradox to take effect thereby cancelling out any potential traffic efficiency gains as more people (individuals) flock to use the driverless cars? Today there are likely many non-drivers who still would like to move from point A to B but don’t because of fear of traffic or safety, inability to park, lack of trust in driving skills/reflexes, lack of legal driving permit, etc. These people, through the lens of Jevon’s paradox in a driverless car world, could suddenly be avid commuters again. We also should consider new uses — e.g. tourism, formal but also informal deliveries (e.g. call work to get your forgotten papers put into the back seat of a car that shows up at your home), or even silly things like booze cruises. I think the modelling for traffic with driverless cars assumes a normal projection of traffic and population rather than anticipating new uses in what is essentially a complex adaptive system.
I think you are describing light rail?
Not sure I understand how this describes light rail. The driverless car users who I describe would not use light rail today because it is not as convenient as they would like it to be and much too public. Even today’s cars have too many drawbacks. Don’t get me wrong, I think we humans are much too quick to value convenience over all other benefits (see the willingness to give personal data in return for using many apps). I agree with many comments from the previous article in this series that point out that adequate investment in mass transport would be a much more efficient way to bring many of the same benefits of driverless cars. But investors sure like “disruptive” over social good.
Today there are likely many non-drivers who still would like to move from point A to B but don’t because of fear of traffic or safety, inability to park, lack of trust in driving skills/reflexes, lack of legal driving permit, etc.
Don’t take it the wrong way, but many (most?) ppl in the US are just uninformed about the concept well executed light rail
This gets it pretty close.
> all I’m seeing in this networked vehicles trend is another kind of rent-seeking through forced obsolescence,
You say that like it’s a bad thing!
As a public utility, huge efficiencies and conveniences could be achieved, but that defies Rule 1.
” huge efficiencies and conveniences could be achieved”
facts not in evidence. You have to prove that, not just declare it to be true, with a HUGE added…because of course….
I agree strongly! No matter how you automate the car, a few hundred thousand people funnel into San Francisco every morning and a few hundred thousand have to get out of the city each evening. It doesn’t matter who or what is in control of the car, the congestion is going to be there. I discount the benefits of “networked” cars because of the hacking concerns – even Google’s cars aren’t networked for just this reason IIRC.
Flexible work schedules? Multiple riders? Telecommuting? They all help defeat congestion but NONE of them require self-driving cars.
mass transit too.
Isnt the most obvious solution for SanFran more multifamily residential properties IN SanFran?
Or close to where they work, say in Mountain View.
I could see automated cars combining with a smart phone app for car-pooling. Rather than having a regular car pool, you’d set a time to leave (or use the app to flag down a ride Uber/Lyft-style). The network would then route a car to your door.
Problems? Heck, yes! Imagine a DDoS attack during rush hour, intermittent GPS signal jamming, etc.,etc. and so forth. Sort of a cyber version of the protesters who chain themselves to subway cars, or block bridges. Not to mention the idea of a self-driving truck full of valuables getting hacked by thieves and re-routed.
On the other hand, an automated local shuttle service taking seniors and/or disabled people on errands has a certain appeal.
> an automated local shuttle service taking seniors and/or disabled people on errands has a certain appeal.
Ok, sure, but what’s stopping us from doing this right now? A $12/hr person to drive the shuttle? If it’s not happening now it won’t happen with automated vehicles either.
And you’ll need a human on the bus to take charge if a passenger has a breathing episode or falls, now won’t you?
I’ve seen a lot of Universities doing work on this front, showing how centralized route planning can optimize traffic flow by distributing traffic in a way that maximizes aggregate throughput even if not every vehicle takes the shortest possible route in the absence of traffic. It’s something that’s easy to simulate, looks good in publications and doesn’t require too much of a capital investment for cash-strapped Universities, but OEMs aren’t too inclined to pick up on it because there’s not much of a way to monetize it, better to add extra cup-holders or whatever for the people who are going to shell out $100k for your particular logo.
The same is true for ad-hoc wireless networking to optimize traffic flow, which can be done through things like reducing shock waves by broadcasting acceleration / braking information faster than radars or other sensors can detect it. It would require a lot of standardization and collaboration between manufacturers and it’s not clear how any individual OEM makes money on it, especially early adopters.
So basically it can be done, but it would be a lot more effective to minimize all those single-occupant vehicles like you say.
Does that centralized route planning account for places you might not want to go? A lot of high-crime areas are remarkably close to main thoroughfares and upscale business areas. Who is responsible if your route planning, in the name of total throughput, takes you someplace where actually stopping at a stop sign or red light is the last thing you want to do?
Thats called urban adventure! Ppl should pay more for the routepkanning subscription!
A few years ago I experienced a crude version of this while using GPS to navigate from my hotel near Disneyland to the La Brea Tar Pits. Nothing bad happened but I did see some of the less glamorous parts of LA.
La Brea Tar Pits
One of the few places in LA I have any interest in. An amazing place. Once I was there I saw bitumen oozing out of a sidewalk crack! HA!
Not to mention greenhouse gas emissions
My great grand-uncle had an early form of networked vehicle that retuned to base.
It was in the guise of a draught horse that pulled his ice wagon on his saloon route.
As the uncle took progressively longer and longer on his stops quenching his thirst, the horse was known to eventually abandon him and walk back to his barn to get his oats and a rubdown. Algo that!
> However, part of the sales pitch for network vehicles is that the vehicles return-to-base and so much less parking would be needed overall. But–that would mean the networked vehicles actually spend more time in the traffic, increasing the overall traffic volume.
Exactly. When self-driving cars are discussed in tech forums I see people gushing over fantasies of being able to send their car to pick up their kids from school, avoid parking fees by dropping them at work then driving home or just constantly circling the road, etc. It seems to me that would make traffic even worse, since now we have countless zero occupancy cars zipping around, contending with the single occupancy ones which create enough problems as it is. Can you imagine how frustrating it would be to be stuck in traffic next to cars with no one in them?
It seems to me that putting more cars on the road isn’t going to solve anything. What we need is public transportation.
I can’t even read the comments on tech sites anymore when it comes to this topic. If you express any skepticism (software certification is a big one with because I work in aerospace), you get called a Luddite by people who would do well to look up the actual meaning of that word.
The other thing they do is disparage human drivers to ridiculous extremes. If people were as bad as they imply, my daily commute would resemble Mad Max: Fury Road.
The Prometheus Project, which was an EU project in the 1980’s to develop driverless cars was primarily intended as a method of improving the capacity of existing road systems. From memory (I had lectures about it in the dim and distant past), simulations indicated that self driving ‘chains’ of vehicles could increase the capacity of a highway by about 20%. The savings comes primarily from maintaining a constant steady speed (counterintuitively, lower speeds are more efficient in congested road systems) and by eliminating the randomness of much traffic movements.
I suspect many of the benefits are exaggerated, not least because road traffic models have notorious flaws which have only been ironed out in the past decade or so. Your Cairo experience reminds me of what was once called the Athens Paradox, whereby traffic movement in chaotic Athens was significantly better than computer models anticipated. Vast sums of money have been wasted in urban traffic schemes to ‘improve’ road networks that were not actually all that flawed.
However, if you sit and watch a traffic junction for any length of time, you will see how an idealised self driving system if (and only if) all the vehicles are part of it, would be more efficient. In regular traffic, every little error or selfish decision by a driver gets magnified through the system. One person a little slow to get moving can mean three or four cars don’t get through a green light who should have. If the sort of self driving systems which were planned in the 1980’s had ever been implemented, you would have had traffic sorting itself into relatively slow moving ‘chains’, all carefully integrated with light controlled junctions. It would be more efficient and probably less noisy, as car movements would be more regular. Whether it was ever realistically achievable is a different matter altogether. The traffic engineers I knew in the 1980’s were quite convinced this type of system would be widely used by the dawn of the Millennium.
This was before my time, but I work on similar systems today and it is very difficult, especially when it becomes a multi-brand exercise and different teams of engineers are implementing their own methods.
How much can I trust the acceleration/braking signals of the guy in front of me? There’s no standard for that. If I’m using his GPS fix, how much can I trust that? How do we maintain the formation in an safe way when we’re passing onramps / offramps? How do we deal with people cutting in the middle of the formation, or even detecting that it’s happening and sharing the information with others? How do you deal with vehicles that don’t have this deployed?
These aren’t impossible technical problems but most importantly, why should I be the first one to implement this? I’ll be the one paying the highest costs to sub-suppliers and there’s no benefit to being an early adopter for such a system.
Yeah, the fact that people are still looking to technology ex machina to solve all problems shows the main problem: that we need to learn to move away from cars and to proven solutions like mass transit, liveable sustainable cities etc.
Well . . . “we” weren’t looking for such smart car solutions. “They” were suddenly looking to flood our brains with “smart car” in order to change the subject and hijack the discussion over to “smart cars”. To prevent any discussion of bringing back America’s good old network of trains, trolleys and streetcars from taking hold in peoples’ minds.
If I ever have to buy a car, it will be the dummest car I can find.
My mother has an 8 year old Buick which has very few miles on it. If there’s any way I can mothball it, I intend to.
To Different Clue: You beat me to it. It seems the purpose (at least in part) is to find ways to keep the microprocessor industry at full employment for the foreseeable future and not because there is a burning need or a better mouse trap.
As a society, we seem unable to adopt “simple, rugged, proven” solutions for anything. This brings single payer to mind.
It also occurred to me that if we used to have “autonomous vehicles” that would take you where you wanted to go with a word of command, after some training; they were called horses.
Of course, there’s the manure problem, a rather neat illustration of Fleming’s idea that mixing tends to be best done centrally, and sorting (as of waste) locally. And I’m enjoying the idea of the Silicon Valley brainiacs being tasked with handling manure instead of developing the next Internet of Things, except for two-ton moving objects, instead of thermostats and routers.
Heh! When the police horses clop by on our street and drop a load of fertilizer, my neighbour and I run out to scoop it up and throw it in our composters.
We are not so keen on the waste products of cars.
> What I would like is an explanation of how networked vehicles would necessarily ease urban traffic conditions.
You’re not going to get it from this post, because the people who know the tech, as opposed to the Silicon Valley crooks and grifters talking their books, say that autonomous vehicles aren’t happening in our lifetimes.
I see the rest of your comment is a strong takedown of the “easing traffic congestion” concept, but to me that would be like writing a post that showed that “electricity so cheap it won’t be metered” (the sales pitch for nukes, back in the day) would lead to unwanted side effects. Event A won’t happen, so why worry that Event A might not cause B?
NOTE Another facet of Shladover’s expertise is networked vehicles; I believe we’re still working on the standards.
I think the solution is the evolution of networked cars to drive in isolated, dedicated traffic lanes. These cars could be physically networked, and increased in capacity; call them ‘smart’ mini-vans. To increase efficiency, they would stop and start in a coordinated fashion at pre-defined intervals. At this point individual ownership would no longer be logical, and the cars or individual seats would be rented on an as-needed basis. Further, the problem of parking would be eliminated because the cars could simply remain in service. At the peak of evolution these tightly networked ‘cars’ would be indistinguishable from the technology still visible in some US cities known as the train or subway. Because they would be self-driving cars, however, they would not have the stigma of public transportation, and would thus not represent an attack on core ‘American Values’.
“[Conditional Automation]: Level 3 is now where the technology builds in a few seconds of margin so that the driver may not have to take over immediately but maybe within, say, 5 seconds after a failure has occurred. ”
Hard to imagine any conditions where this would apply. FIVE seconds is an eternity in reaction time. Most unimpaired drivers can react and bring a vehicle to a full-stop in under 5 seconds (at road-legal speeds.)
Our two cars are 10 and 5 years old. I hope to have them both for 5-10 more years (longer if I can).
The average age of cars on the road is over 11 years. So who is going to be buying these auto-autos when they have a perfectly good car in the garage?
Several answers to that :
A) people who have cars 15 and 20 years old, like you in 10 years…
B) people that have cars which can still work, but cannot drive because of developing infirmities (a.k.a. Old age)
C) people who never had a vehicle before, especially if the don’t have a driver’s license
D) taxi Operators
Other models include rentals, and fleets, where you simply hail your car, instead of owning one.
“Simply.” Thereby hangs a tale.
“Here are a few scenarios that highlight limitations of current self-driving car technologies: The first is operation in bad weather including standing water on roadways, drizzling rain, sudden downpours , and snow”
Humans don’t do well in any of those situations either, this a red herring.
“Google X has advertised that its cars have driven 2 million miles accident free, and while I applaud this achievement, New York taxi cabs drive two million miles in a day an a half.”
Judging from the taxi drivers I have seen, that day and a half include many accidents and near accidents. This is an argument for autonomous vehicles.
Automated driving has already started in a few places around the world and set to expand tremendously by 2020 as more competitors enter and more data is collected and analysed. Car ownership is not going anywhere but driverless taxis and trucks will be everywhere sooner than most realise.
Bullshit. “Standing water in roadways” and “drizzing” are trivial conditions. And please tell me how many accidents result from “sudden downpours”. I’ve lived in two tornado belts (Dayton, Ohio and Birmingham, Al) and can’t recall one mentioned in the local papers in all the years I lived there. People don’t go out once they’ve started, so traffic density falls, and having been in them often, drivers slow way down and in really bad ones, pull off the road. And as for snow, despite driving only for about a year in seriously snowbound area, I can snap out of skids in snow rapidly. Most people who’ve driven any length of time in snow get to be good or extremely good at it.
I thought the problem with standing water/snow etc is the proposal to have auto vehicles track by road lines which are obscured by debris?
You’ve got road lines?!
There are many techniques for automatic driving software to track a path: locking on well-known features (e.g. road markers), following textures, discovering contours and shapes, etc, using cameras, infrared sensors, radars, etc.
Correspondingly, there are many problems popping up when trying to remain on a lane:
a) Water will reflect objects on and beside the road, and make recognition of the lane more difficult.
b) Water at night will reflect lights on preceding and of incoming vehicles, and make assessments of distances of moving objects tricky.
c) Snow, especially outside towns, may hide the difference between road and off-road.
d) Drizzle and fog severely reduce the effective range of infrared sensors (one of the methods used for automated vehicle driving). Actually, in what is classified as ICAO category 3a and 3b fogs, infrared cameras typically have no better than, and often lower detection range as the naked eye. Heavy pollution is also detrimental to infrared-based sensors.
Apart from that, parked vehicles, roadworks, lights from nearby buildings (e.g. shops), temporary detours, dirt on the road (spillage, sand, earth), sudden changes in lane patterns (e.g. when a portion of a road is being rehabilitated) all massively complicate lane detection algorithms.
Individual techniques are typically very good at handling certain classes of scenarios and are weak in in others. Combining several techniques improves the reliability of automated driving significantly. However, it also requires a battery of various sensors, and massively more computing power, because all data acquisition, analysis, comparison and reconciliation of various lane detection algorithms, and reacting on vehicle actuators must take place in real-time. Which means expensive vehicles.
For people (technically) interested, there is a survey of approaches in
S. Yenikaya, G. Yenikaya, E. Düven: Keeping the vehicle on the road — a survey of on-road lane detection systems, ACM Computing Surveys, v. 46 nr. 1, October 2013.
Finally, have a look on Youtube about driving in Russia. You will see plenty of crazy drivers, trucks shedding their load, trucks flipping on the side, tanks crossing streets, packs of wolves running on the road, mud slides — you name it. Genuine level 5 automated driving vehicles will have to deal with all that.
If the solution for bad weather condition is to drive very slowly or just stop by the side of the road, I am quite confident that autonomous vehicle will be able to implement it. Actually, they will be probably better at it than humans because their judgement will not be biased by a “I can make it home” mentality.
” they will be probably better at it than humans because their judgement will not be biased”
Do you have any evidence of this? I call BS, on both the idea that they will probably be better, and they they will ever be able to drive in any sort of adverse weather.
Speculating doesn’t count as evidence.
Taxis accident in NYC is about 20 per day. See http://www.nyc.gov/html/tlc/downloads/excel/vz_tlc_lic_veh_crash_data_ll_31_2016.xlsx
So one day and a half is 30 accident vs 0 for Google… go figure.
Until you have evidence that Google has had automated cars driving in NYC for a day, then I might be able to compare the taxi drivers with the software. Oh and read the quote:
“Google X has advertised that its cars have driven 2 million miles accident free, and while I applaud this achievement, New York taxi cabs drive two million miles in a day an a half.”
I can advertise that cow chips will cure corporations of their own arrogance, but that doesn’t make it true. I am glad that we are holding up advertising as evidence.
> Judging from the taxi drivers I have seen, that day and a half include many accidents and near accidents. This is an argument for autonomous vehicles.
The word “testing” isn’t in your vocabulary, then? That was the context of the statistic.
“Situational awareness” is not something you can summon with a beep.
Orys and Roland
Beyond the difficulty of having to orient yourself to a rapidly developing emergency situation, this rules out one of the bro-ier selling points for automated cars – no DUIs. I would bet that this will make possibly having to drive drunk much more palatable for people who would normally never even consider doing it.
On an unrelated point, I’m in Boston and have been seeing announcements that the city will be welcoming its first automated car(s)in the near future. I don’t ever recall agreeing to serve as a guinea pig. I can’t even begin to describe how mad I get at the idea of someone I care about, including me, being the bicyclist that helps google learn a very valuable lesson. I may have missed the vote that was taken, but I haven’t heard anything about the populations of these test cities having any say at all in the matter.
> no DUIs
DUI deaths are disproportionately in rural states. But all the initial proposals for self-driving cars involve urban use. So yet again one of the main putative benefits turns out to be handwaving.
> I haven’t heard anything about the populations of these test cities having any say at all in the matter.
Well, of course not. “Innovation” is good in itself, no need to put it to a vote; see Thomas Frank on, as it happens, Boston, “spiritual home of the professional class.” And no doubt the innovators take the city official responsible for the project out to a very good lunch.
> the bicyclist that helps google learn a very valuable lesson
Heh. I wonder what the bicyclist community thinks of autonomous vehicles? They can be tough political players, is my recollection. Google says it’s on it, but we already know their testing sucks… In the article, the bicyclists are riding around the car in a circle. That doesn’t seem very realistic to me.
That’s a really interesting point about DUIs being also much a rural problem. As to the bike problem, I wonder if a car company will eventually decide the goodwill would be worth the cost and just offer a free transponder or some sort of add-on to all cyclists that makes bikes very easy for cars to pick up on.
The answer is always more technology; in the form of self driving bicycles.
What I would like is an explanation of how networked vehicles would necessarily ease urban traffic conditions.
How will all those autonomous vehicles deal with me cutting them off in a 1953 Austin Healey as I drive around the man standing in traffic eating his ham sandwich??
Your ’53 would only be allowed on the original Route 66 – next to and below the future.
well, my house is abt 7 miles from rt 66 , so I’ll have to run a few JohnnyCabs off the road to get there!
I might have to give the computer better odds. You might be able to assume that the computer’s electrical system *might* work once in a while.
That’s a long shot bet with an austin.
And it seems to me that these Southern Californians haven’t given much consideration to snow. I imagine that the current state of the art has difficulty when the lane marking are obscured, there is snow obscuring vision, and traction is minimized. That said, I’d really like my parents to have all the driver assist technologies possible.
In addition to the lane detect cameras, all the other graphical processors are going to have a tough time with snow and so will Lidar arrays, so the redundancy of sensors required for building a picture of the environment is majorly crippled and I haven’t seen any good technical solutions for addressing this.
CA does not have nice roads. Wealthy cities might but roads are constantly under construction. We are worse than OR or WA and our frost date isn’t until Nov 15 and we already have snow. We are a mountainous area with leaves, moss, standing water not to mention single lane, land slides and sloughing.
The cars seem to be best suited to large cities. Instead of lines, the cars could possibly track off of markers placed high on existing polls? If traffic is so heavy and individual vehicle can’t interact with lines they require networking. Once they are networked? That is nothing more than public buses. So yeah, it’s rent seeking.
How will those self-driving cars fare in an earthquake?
“stay between the lines on the bridge”
But the bridge is falling!
A great metaphor for our times!
> I’d really like my parents to have all the driver assist technologies possible.
So would I. That is different in nature from a (Level 5) self-driving car. More from Brad Templeton’s piece:
Driver assist is “capabilities and features.” But a company that does that has a valuation like GM’s or Delcos. A company that does a Level 5 vehiicle — or can plausibly claim to long enough for insiders to cash in — has a valuation like a Silicon Valley unicorn. So that’s what crooks like Kalanick and grifters like Musk say. They say what they’re going to say. They frame the development of capabilities and features as steps in the inevitable progress to Level 5. But it ain’t so.
Nvidia, a graphics and parallel processing hardware company, recently put out a video on a “self taught” driverless car based around an infrastructure called “DaveNet.”
What is interesting about this approach is that it is about teaching AI to mimic human behavior instead of trying to account for all the rules. The problem is that not all situations can be accounted for – but for that matter, people are barrelling towards Level 5, even if the infrastructure never actually gets there.
Sadly, these AI approaches create black boxes that offer zero visibility into how the car might be making decisions. At some point, you will need to “prove” that the system will behave in a consistent and safe manor. That can’t happen with the current state of trainable AI.
Out of interest, how do you prove that a human driver will behave in a consistent and safe manner?
Excellent summary of the technology development issue here. As discussed last week remember that Uber and Lyft get zero benefit from self-driving cars until NHTSA’s Level 4 [SEA’s Level 5] is fully achieved, unlike the auto manufacturers who have the potential to benefit from each incremental step. But 90% of the hype about driverless cars comes from the owners of these “ridesharing” companies or (more importantly) the mainstream media people who have drunk the Silicon Valley kool-aid and report its favored narrative as established fact.
Another way to understand/explain this is to look at railroads. In theory the obstacles to automation would be a tiny fraction of what automakers face and returns more immediate. Railways are nearly closed systems and the railway operators control every aspect of the right of way, unlike roadways which are totally open to pedestrians, cyclists and people driving non-automated cars (or driving drunk) and where no one entity could possibly ensure that the roadway and communication standards needed to support full automation. The costs of major railway accidents are huge and every operator has a well-developed safety culture and proactively manages these issues. But automated rail lines are strictly limited to 100% isolated people-mover systems (which are really just horizontal elevators) and some remote control of switch engines in isolated freight yards. The rail industry is facing a major crisis because Congress mandated (in 2008) the installation of Positive Train Control by 2015, but the technology just isn’t there yet, the current costs are astronomical compared to medium-term benefits, and it is very difficult to roll out gradually. The safety/human error problems automation is supposed to be solving only arise when the network and the traffic mix become much more complex. To get a big safety benefit you have to automate the entire NYC subway system or the majority of the Norfolk Southern railroad. And absolutely no one thinks you ever get to the level of safety where you can eliminate the train drivers. But the “journalists” who faithfully report Uber/Lyft claims that completely driverless cabs are just a few years away never bother to ask the railroad people who have been working on (comparatively) much simpler issues why it is taking so long and costing so much.
It is worthwhile for society to be investing billions to find ways to make railroads and highways safer. The problem is that the investor class/mainstream media agreed narrative that “new technologies will rapidly solve all our problems, but please don’t make me explain exactly how this will be possible or ask me to consider decades of experience showing these challenges are actually very difficult.” Thus Congress “solves” the problem of freight railroad accidents by implementing non-existent technology and Uber investors create $68 billion of corporate value from the same source.
There’s a big difference in difficulty between building a buggy Android app designed to surveil your customers and serve them ads vs the real world challenge of moving around a two ton steel box filled with lives. The consumer tech industry has no track record of providing technology that’s safe enough to perform without human assistance. These tech execs are talking up their share prices. It’s difficult to believe the amount of hype behind this technology. In the software world we refer to vaporware, a product designed to solve your every problem except for the fact it doesn’t exist.
I worry that once people figure it out it’s going to discredit the technology industry for years. Right now, people who have no clue about the technology have bought into the hype that fully automated cars are 5 years out. These are the same people who believe their “smart” phones represent some sort of revolution, a consumer product that’s designed to become obsolete every 18 months and unbelievably insecure. I suspect that the solution will be to redefine “self-driving” to mean something more like Level 3 than Level 5 on the chart. As Obama has taught us, better marketing is the solution to every problem.
Great write up, Lambert.
Good point Hubert.
It would seem that railroads would provide the ideal platform for level 4 or 5 automation due to extensively standardized “roadways” & sophisticated scheduling schemes. Yet there was still an incident involving extensive property damage, injuries & at least 1 fatality in N.J. last week.
Also, as Qrys points out, level 3 automation is highly questionable; and the N.J. wreck may have made that point. Did the automated system malfunction & the human driver didn’t notice in time to override?
Praise from Sir Hubert is praise indeed.
Perhaps I made the valuation point at the end too gently? I thought saying that Uber should be valued like Delco would cause howls of outrage, but apparently not…
I think they should start with a self-driving hearse: it only has to go from the funeral home to the cemetery, and the passenger is already dead!
A disaster evacuation is another scenario that springs to mind. It’s the same exact problem electric vehicles are going to have.
Huh? If power is out GA stations do work either. I I disaster you are stuck with what you have in the ‘tank’ regardless of whether it’s gas or electrons.
Stupid phone. Gas stations do NOT work without power
Whenever I hear about this level of automation I always think back to the 2008 crash of a BA 777 at Heathrow. Both engines failed at 720 feet altitude, 2 miles from the runway (3.5-4.0 degree glide slope). The question is: without the pilots taking over manual control @150 feet, reducing flap to stretch the glide, and just squeaking over the perimeter fence (IIRC the pilots had to `bump’ over it) would the automatic systems have actually made it ?
Which reminds me of that other problem with highly automated airplanes: The atrophy of pilots’ flying skills. Sounds like those BA 777 pilots still knew how to fly well enough to handle dual engine failure.
Whenever there is some combination of snow, sleet, or fog in the high sierras,
all the “giggle-wagons” are stuck, but cars with drivers get through, if not easily.
Yuppie-nerds and their fantasies are laugh-worthy if nothing else.
The unnamed city will be Autonomous Taxiville, Kansas, population zero. It will have a single loop road that doesn’t connect to anything. There will be a taxi rank for the fleet of one (1) autonomous taxi, which will remain stationary unless passengers appear. When asked how many passengers the taxi has carried, the company will decline to reveal the exact figure for commercial reasons, but state that it is a number between zero and one million.
Reminds me of this city:
Indeed! Funded by the State of New Mexico, so that’s quite a grift.
I generally agree with the sentiments in this thread, and the comments about the need for open standards for validating the software are especially trenchant.
However, how many drivers do you know have Level 5 skill? From my daily commute I’d say it’s a fairly small percentage. But it might be my confirmation bias showing.
I took on a part time delivery job this summer, and I was horrified by the behavior of the other drivers on the road. Everyone makes mistakes sometimes, but spending 20+ hours on the road every week made it much more obvious about how many reckless, aggressive, and distracted drivers there are. I understand that the technology still needs work, but in the long run, I’d be more comfortable with the dispassionate calculations of a computer rather than the selfish, raging, distracted decisions of a human.
I think there may be a generational component to this discussion too. I know of very few people who actually like driving. In conversations among my friends and family, everyone (with the exception of my brother-in-law who just loves cars) is eagerly anticipating driverless cars. It somewhat reminds me of the conversations about using ATMs. My grandfather was horrified by the idea, and said that he’d never use a machine. My mom defaults to an in-person teller, but will use one when necessary. The only time I go to the credit union is if I need to sign papers for a loan (and the last couple of times that was done electronically too).
I have four kids, aging parents/in-laws, a mother that’s a horrible driver, and find driving to be very stressful. I would pay significant cash to have a safe, driver-less car, even if it could only be used in certain conditions or in-town. Maybe the technology is over-hyped, but I think we’ll eventually get there, given the huge demand for this type of service.
The technology exists today – it’s called “training bus drivers”.
> I know of very few people who actually like driving.
That’s quite distinct from how cars were orginally sold (and I’m not sure it’s true for teenagers, even today). I also think that when people understand they’re getting into a machine that has the power to decide to kill them if an algorithm thinks that’s a good idea, they may think differently.
I would imagine that “the commute” takes a lot of the fun out of driving (and is a terrible stressor). However, AFAIK, a lot of “the commute” is from the suburbs into the city (traffic pattern experts please correct). If so:
1) The suburbs are increasingly becoming dumping grounds for poor people as the cities gentrify. Hence, the making the commute in from the suburbs won’t be the first priority, even if Level 5 self-driving cars somehow became a reality
2) The flip side is that many scenarios for self-driving cars involve them being rolled out in the (gentrified) cities first — that is, where the benefit to society as a whole is least, but the personal comfort of the 10% and the 1% is maximized.
It’s interesting that discussions of the life-saving potential of self-driving car technology never mention the life-saving potential of actually-existing technologies: public transportation. Replace highways with rail and traffic deaths go down dramatically. Implement car-free downtown and convert roads to bike lanes, as Oslo is moving towards. Or even put an exorbitant toll on low-occupancy vehicles using roads to encourage bus-riding. All technologies that currently exist, but that don’t allow for rent extraction.
I have no complaint with bike lanes, low-occupancy tolls, or car-free downtown. However, the “just use public transportation!!” response always irks me a bit.
Some people are exposed to significant risk and harassment in public places and/or transportation. When I’m driving in my car, I’m not going to get cat-called, bothered for a date, or flashed.
As a parent, I don’t have to worry about disturbing others if I’ve tempted fate by doing grocery shopping late and my kid is crying and needs a nap, or if my kids are having a bad day and arguing. I have kids in various after school activities that occur in different places at different times. I’m not going to send my 6 and 9 year old girls on a bus by themselves, and then have them sit in a park alone for an hour because the bus only runs hourly.
I can relax and have a few moments of quiet in my car, rather than listening to someone else’s loud music, one-sided cell phone conversation, hallucinations, or heated argument with his/her partner.
People also vary in temperament (i.e. introvert versus extrovert). Being in a public setting is inherently stressful to me, as I can’t just block out everything that’s going on around me.
If you can create a “private,” single-podded public transportation, I’ll be the first person to sign up. I agree with the goal (reduced energy use, cleaner air, etc.). We’ll reach that goal more quickly if we can also address the needs and concerns of the those who do not currently use public transportation as their preferred transportation mode.
Lucky you, living in a society that actively promotes destructive lifestyles.
Fortunately, these are also problems that other countries have solved.
Women only cars in public transportation
In Japan, children as young as 2-3 years old run neighborhood errands for their parents. They use the subway alone to get to school from the age of 6. Maybe parents in other countries could stand to be a bit less paranoid and worried about being judged.
Finally, regarding an introvert’s “need” to be alone, I myself am an introvert and can testify that a pair of headphones goes a long way toward blocking out distractions. Feeling anxious about appearing in public might be another matter, though.
I should also investigate international driving fatality rates. If the statistics are anything like health care statistics, we’ll have a very high death rate combined with incredible costs and lousy systems.
Of course, things like driver training and licensing are gummint things, therefore against Silicon Valley glibertarian orthodoxy.
From the Cummings testimony, I thought this bit was eyebrow raising:
It is not uncommon in many parts of the country for people to drive with GPS jammers in their trunks to make sure no one knows where they are, which is very disruptive to other nearby cars relying on GPS .
The normal mode for GPS is that satellites transmit and devices like smartphones only receive the signals (and then do Einstein-esque math, involving speed of light, to work out position). So jamming will disrupt nearby devices, and also prevent you from knowing where you are. But it won’t stop license plate trackers or “eye in the sky” satellites from tracking your movements. So why do it?
Is this really common? The info here suggests it is illegal. Also, if you are trying to hide, operating an unlicensed transmitter seems like it will undermine that goal.
I think maybe what she is getting at is trying to prevent a device from tattling on you to by retransmitting your GPS position over cellular… but if worried about that, consider putting the phone in a lead lined box.
1. I believe the license plate readers operate only at certain points, like toll booths. They aren’t universal like the surveillance cameras in Central London. So they can infer your route but not prove out all the details
2. Some of the jammers may be visitors renting cars.
3. There is a difference between being a person who is actively surveilled (where the satellites might be watching) v. a person who is concerned that all their movement data is being hoovered up along with everyone else’s so it could easily be mined later if they were to become a person of interest.
4. If the unlicensed transmitter knocks out nearby cars too, it probably can’t be tied to one car. So unless they are caught with it in the car and running, the operators would seem to be able to get away with it.
However it would seem a lot easier to take the batteries out of their devices or put them in foil or something else that would block the GPS signaling.
GPS jammers are fairly easily located if the right tools are present.
In fact there is an app for that (only runs on Android and it not in the app store for general distribution but does exist). It does require a number of people to be using it in the same location. The idea is that it runs in the background and is networked.
Lots of police cars have license plate readers. I see them all the time.
Some private corporations run license plate readers also.
But police cars are not ubiquitous. I don’t even see them all that often in NYC and they aren’t sighting them 360 degrees (and there are cars around so they can’t catch all that many). So you are confirming that car locations are sighted only at certain points, which means the route in between is subject to question (save maybe for multiple sightings along the same interstate), while GPS tracking provides a complete record.
Yves, FYI, it appears that license plate readers, like the rest of the Panopticon, are fixed, ,mobile, flying, ubiquitous and getting more so. Like the rest of the Matrix.
Interesting fairly recent article from the Atlantic, noting how useful these gadgets and their applications have “helped police and lenders target the poor” and a lot of other people too, though the “poor” appears to be the larger set… http://www.theatlantic.com/technology/archive/2016/04/how-license-plate-readers-have-helped-police-and-lenders-target-the-poor/479436/
And this, from 2014: http://spectrum.ieee.org/cars-that-think/transportation/sensors/privacy-concerns-grow-as-us-police-departments-turn-to-license-plate-readers Even back then, it seems 70% of local and state cop cars have readers installed and in use. Too bad the dash cams and body cams are not as ubiquitous and, ah, constantly operational…
Gee, no possible way that the “Fusion Centers” include Hoovers dedicated to watching and storing all this information?
Anyone run across the mission statements and specs for how all this stuff is to be used? Like the numerous “mine-resistant vehicles” now operated by the so very terrorist-threatened Pinellas County (FL) Sheriff’s Office?
“Resistance is futile…”
Yves, a comment with a couple of links went to moderation or skunet. A cursory search on “license plate readers in use” indicates that by 2014, maybe 70% of police cars have them, and they are everywhere, mobile, fixed, flying and spying. And that police and “lenders” just love the technology, particularly applicable to “the poor.”
I agree that most of the hype is coming from corporate execs trying to goose their stock price. The really amusing part is the credulity of reporters, economists, and the TED talk brigades. There is a deep need to believe that the driverless cars are right around the corner, perhaps because they really like the idea of driverless cars, or in other cases wanking over the idea of all those out of work cab drivers and truck drivers.
The problems are so obvious to me and I have no technical background. As an additional issue I also fail to see how Uber would profit from a driverless car even if it were possible. At present their workers/drivers are responsible for storage, cleaning, maintenance, insurance and fuel costs. A self driving fleet would incur all these costs and many cars would be idle during non-peak times. This whole thing seems infected with motivated reasoning and very little critical thought.
I think it’s mostly flimflam for the rubes. But the endgame would be to eventually drive normal cars off the road, encourage/force people to subscribe to an oligarchic car service, and then crank up the rates.
With a flat tire, who or which will do the tire-changing?
The car or the passenger?
“You do it, car.”
I think these guys watched too many Jetsons reruns. I, on the other hand, watch too may spy/crime movies. What happens if you get in your auto-auto and the doors lock and you get whisked off to a police station, jail, black site, or underground bunker? Kidnappers can have a field day. Please remind me why we need these things. How about better public transportation as reslez says above.
Self-driving cars….. will they be named Christine? They’ll be possessed by external forces for sure.
I’ve never understood what the end game is for having autonomous vehicles. There are so many unanswered questions regarding insurance, ethical questions described in your previous post on this subject, software problems, and software upgrades. Who’s responsible for the software? Recently, at least two major airlines went down because of software upgrades – will the car suddenly freeze up?
Also, being hooked up to the internet there are endless possibilities for spying on us, tracking our every mile – and yes, more cars on the road! How is this even remotely something people would want? We need more public infrastructure. I’m so tired of hearing that technology can fix our every problem we humans face. It’s nonsense.
Lambert is right – this is just more rent-seeking “entrepreneurship” from these hucksters/unicorns.
The end game is two-fold: 1) feed at the public trough of infrastructure spending; 2) provide a captive advertising audience (“so, you are going to the mall, what store”, “would you like a coupon to visit XYZ while you are there”, “and if you purchase, the ride is 25% off”). What, you thought you would have free time in an autonomous vehicle – it is to laugh.
Think New York City cab video screens but without a (working) mute button.
You can also combine:
1) Cashless society with
2) A rental model for vehicles
3) The FIRE sector completely controls your ability to be mobile (remotely). Not that there’s anything wrong with that.
Assuming that Level 5 vehicles are even possible.
Actually Arthur C Clarke in Imperial earth previewed where we might be going. In this world, freeways were automated (because they were controlled access places) and for example you had to manipulate the controls on a long driveway. I suspect that such vehicles will never work without operator input on many jeep roads (see youtube for examples of such roads say around Ouray Co).
The dispute about ethics neglects that if the car refuses to go without your seat belt being fastened, between the seat belt and the air bags about the only situation that could arise with the decision is a head on collision. But then if both vehicles jam their anti-lock brakes on the closing speed is reduced and thus the damage. Now the car might be programed to do what is suggested in Drivers ed, to sideswipe a car rather than head on it (or the particular example, I recall is brakes failing on a mountside) where the best option was to sideswipe the hillside.
i’m old enough to remember the “kitchen of the future” widely touted in the fifties…they’d have a dress-up housewife sitting in middle of the kitchen as appliances and cooking are all run from a control panel, with robots running around performing the menial tasks…
see how much of this came to pass: http://www.youtube.com/watch?v=TiACOLuYlJ4
Herman Herman? OK, then…
Lambert, a fine analysis, but unnecessary.
It’s the old Batra Bezzle, updated with robots. “My predictions were wrong, but you bought my book, and I have have your money, so F-you too!!”
Elon Musk? The guy who took a break from actual and financial auto-fellatio to issue a plan for colonizing Mars?
Travis Kalanick? The guy who wants people to believe that 35 tons of tractor-trailer hurtling down the freeway at night with the operator asleep behind the seats is safe?
Are we done here?
Uber’s entire investment thesis rests on driverless cars working not too far in the future. There are tons of other true believers. Who is appointed Secretary of Transportation in the coming Cabinet will matter a great deal.
The former NHTSA administrator under Obama, David L. Strickland, is now working a gig at a lobbying firm on behalf of an industry consortium including Uber that deals with vehicle automation.
But will Uber be able to find driving AIs smart enough to drive but gullible enough to work for them?
The very concept of self-driving cars is such a typically American bit of idiocy, born of the techno-narcissists of Silicon Valley. America has plenty of problems with transportation, problems that simply changing the nature of the cars won’t fundamentally fix because cars are part of the problem.
If we’re going to pursue technology that takes away the occupants ability to control the vehicle, why not just be sensible and improve public transport? If Canada or Scandinavia are the ideal for healthcare, Japan should be the ideal for transport: bike and pedestrian focused environments with lots of trains and buses for longer distances. It amazes me how a country we reduced to literal rubble can develop crazy bullet trains but its conqueror, the self-styled Exceptional Nation, can only shrug its shoulders. “We’re not Japan, it simply can’t be done here.”
Of course. But “Can’t Do That” capitalist America is and has always been about maximum profit through maximum waste, i.e., through maximum externalization of costs. That’s the problem with countries evolved from a colonial settler base. It’s hard to break the historical habits derived from 200 years of free lunch on a continental scale, courtesy of indigenous land and African slave labor. The result is the greatest monument to waste in human history.
Turning off that historical road will take a revolution. The PTB here are aware of this, and hence their motto: “Can’t Do That!”
A solution in search of a problem. It just goes to show how desperate our minders are to prevent mass transit.
How long would a journalist stay employed if he didn’t play along with the Silicon Valley stock pump? Ya gotta be a team player right? When the truth comes out about the driverless car scam then if the big boys aren’t out of the stocks in question some big and important rice bowls get broken. No mere journalist can afford to break any big rice bowls.
I am in the industry — also an intermittent reader since 2007, stirred to comment for the first time to add context to this game of automated vehicle futurecasting.
For starters, a correction: NHTSA in its Federal Automated Vehicles Policy issued in September gave up using its own classification of AV systems and adopted SAE’S 0-5 scale.
Next, Dr. Steve Shladover, a key voice in your analysis, is an accomplished researcher of high integrity with a very – to many, a laughably – conservative view of AV deployment milestones. He is, through and t. He may be right, but the same goes for Albert Edwards about the S&P (actually I would give Edwards much better odds). In any case, I find fault with this source, and with others or with the contextualization of their quotes, e.g. Herman Herman’s. CMU is a partner with Uber on current self-driving cars.
Private researchers would agree that “[t]here are fundamental problems that need to be solved,” that cybersecurity issues are safety-critical, and that other basic AV research statements are valid as well. They’re competing and sometimes collaborating to solve these problems Not all, but some of that Silicon Valley cash is going to fund the work of skilled, non-tenure-holding designers of the vehicles of tomorrow.
Returning to the Federal AV Policy, it applies a concept of Operation Design Domain to define the set of parameters within which a proposed (see the new rules!) AV system can safely operate. It is largely these factors, which may include location, traffic, pavement markings, etc. which will determine the limits and expansion of AV deployment across space, time and… driving conditions!
Thanks for the correction on SAE/NHSTA standards; glad I picked the right one!
On “laughably conservative,” no doubt that is what the Silicon Valley engineers think. That strikes me as a, shall we say, sociological phenomenon, rather than an analytical one, the software engineering equivalent of Wall Street’s “I’ll be gone, you’ll be gone.” When this project fails, on to the next one!
There has already been industry / government discussions that touch on some of these things. Driving conditions were one of them. The position of those in the government in at least one of the meeting was that in certain driving conditions the cars would be turned off and not allowed to be used. This meeting took place in DC right after a big snowstorm a year or so ago. At the time DC and MD had requested everyone (with a few exceptions) to stay off the roads. MD at least said they would fine people out driving without a valid reason.
Not to mention the social economic barriers posed by actually existing capitalism, even if all the technical issues were ironed out.
The real potential of self driving cars is to vastly reduce the quantity of cars presently littering the cityscape, spending 95% of their time sitting around doing nothing but taking up space. Cars pooled in fleets and rented on demand could be kept in constant motion, meaning fewer passenger cars required to move the same number of people. Vast stretches of parking space freed for more productive or aesthetic uses.
But one somehow doubts that the auto industry or NADA (National Automotive Dealers Association) – one of the most powerful lobbies in Washington DC – will idly sit by and watch their business model destroyed. They’ll sit on this just as they’ve sat on electric cars, even though technically speaking the entire auto base could have been converted to electric years ago. That is exactly why the traditional auto manufacturers are less than sanguine about the prospects. So, as with Obamacare, the clumsy, wasteful and inefficient compromise will be based on individual ownership, defeating the main advantage of the technology.
> The real potential of self driving cars is to vastly reduce the quantity of cars presently littering the cityscape, spending 95% of their time sitting around doing nothing but taking up space.
I don’t see how that follows. If 60% of the population needs to be at work or school between 7:30 AM and 9 AM, how does the need for cars decrease? Doesn’t that only happen if people somehow end up car pooling? And why does car pooling need automated vehicles? It seems to me the number of cars on the road might actually increase due to people who currently can’t drive suddenly gaining the ability to go out on the road, plus all those people sending their cars on driverless errands they otherwise would have strung together into the same trip, like picking up the kids from piano practice.
If it were me I wouldn’t get in ANY car that automatically required me to cede my rights in terms of being compensated if the machine malfunctioned.
I wish them lots of luck getting past that and outlawing private driving on that basis.
(and I say that as a person whose area is heavily invested in smart cars and that understands that some of the reason these cars are being created is to reduce accidents due to driver problems such as fatigue, distraction, etc etc)
Systems in individual vehicles cannot improve urban traffic flow much, but if the vehicles are networked and controlled together, much improvement becomes possible. You can convoy vehicles at much higher speeds than is now safe. When a light turns green, you can start all the vehicles at once, instead of each one having to wait for the one in front of it to start off. Even if people aren’t looking at their smartphones when the light changes, it takes time for the 10th or 20th car to get to start.
With electric vehicles and with hybrids that are driven by electric motor not by the gasoline engine, it should be possible to make modular cars. So when you commute to work by yourself, you ride in basically a comfy chair with a nice sound system. On the weekend, you combine that into a larger vehicle to go out in the country with your family of four.
What all of this is going to require is a capacity to work together on scales larger than those of corporations. That requires trust. After 50 years of neoliberalism, we have less of both and not enough of either.
They can if Level 5 self-driving cars are possible. But they aren’t.
Whether Level 5 self-driving cars are possible or not, a Level 5 network of mutually communicating cars is a far easier problem technically than building a single non-communicating car that has to deal with current traffic.
We are not even considering that because thanks to neoliberalism, we are socially incapable, although we were probably socially capable when we built the Interstate system.
I think this is an important distinction you’re making that maybe isn’t clear to readers based on comments above. Networked and autonomous are separate ideas.
You can have a fully autonomous vehicle that relies entirely on sensory data to understand where it is and how to make decisions with no data connection to the outside world. This is a relatively trivial engineering problem in closed environments but an incredibly difficult one on public roads.
Connected vehicles can span anything from getting route suggestions based on real-time traffic info to a heads-up advisory on when a traffic light is going to go green to highway convoys where longitudinal (acceleration and braking) are autonomous.
The latter (connectivity) is more resource-effective, but requires collaboration which we’re increasingly crappy at, so we’ve seen a lot more focus on developing expensive sensory and data churning technology that can allow someone to watch a movie in a traffic jam.
Not just “collaboration” but making everyone give up traditional cars all at one. Na ga happen. People need traditional cars for things like dirt roads and off road driving. And a lot of people like to drive. If you think making people give up their gunz is hard, just wait for the explosion if you tell them they have to give up being able to drive a car forever.
I’m not sure I follow your thinking. Connectivity doesn’t exclude the idea of private vehicle ownership. Virtually all modern cars have some sort of telematics system that counts as “connectivity” if you read real-time traffic information and the driver makes some decision based on that. Perhaps the vision of the more extreme Silicon valley types excludes private car ownership, but there’s a whole spectrum of possibilities in between.
I’m in favor of implementing ad-hoc or centralized networking technologies to give priority traffic light signaling for buses/trams. There’s also a cool pilot deployed in Aarhus where RFID tags in bicycle wheels make traffic lights aware of bicyclists. Those are both examples of connectivity that are easily implementable and give tangible benefits. These are things that can co-exist with privately-owned, gas guzzling, unconnected vehicles if we so choose.
Have you been to California? Have you seen how many old vehicles (and some very well maintained) there are on the road in parts of the US with no winter, hence no salt on the undercarriage to shorten vehicle life greatly?
I can tell you from what my father went through in the paper industry with paper mills, and the same issue comes up with trying to make homes really energy efficient: retrofitting costs way more than building new. You are going to impose large costs on consumers with a switchover, and people with older cars will be forced to buy new ones. For people on low incomes, it’s an enormous burden. And I’d be pissed as hell. I like to drive and do not want to give up my control over a vehicle, particularly given the huge lack of reliability in software.
Yes I lived in California for several years and understand that there are far more older vehicles in the US, but the point I’m trying to make is that “vehicle connectivity” is a broad concept that doesn’t require any specific critical mass of road users or necessarily imply retrofitting of old vehicles.
If I get a real-time traffic advisory from my telematics system telling me to take an alternate route to work, that’s a low level of connectivity. If traffic lights deploy a Green Light Optimal Speed Advisory (GLOSA) system, only users with a receiver on board will decode the information.
The more users who adopt such systems the better the efficiency gains can theoretically be with effective implementation, but it can be entirely independent of automation of the driving functions or vehicle ownership considerations. There’s currently a proposed legislation to require the implementation of “Vehicle-to-Vehicle” communication on new vehicles in the US, but it’s all on hold because of the election.
It’s all somewhat academic for me though, I live in Europe and only own a bicycle :)
I just thought of one situation that worries me.
I am in a line of cars in a right lane. There is a guy in the left lane that is hemming me in from pulling into that lane so I can pass the guy in front of me. I accelerate into the space in front of me so that I can get far enough ahead of the guy on my left to pull into the left lane. Just as I am about to use up the last bit of space in my lane to pull into the left lane, my car decides it isn’t going to make it. So, just as I get partially in front of the guy to my left, my autonomous car decides to put on the brakes.
How can I possibly drive a car that might work against me at the most inopportune moments?
That’s why you need my soon to be patented Change Lanes Onto Autonomous Knuckleheads (CLOAK) device. With transmitters mounted discretely in the front wheel wells, the CLOAK will send signals to the sensors on the hemming knucklehead that makes it think you are already in the left lane. That car will then put on the brakes, and you can change lanes freely.
Wheeeee! This autonomous tech is going to present so many opportunities for pranks.
I personally prefer the “jet-pack” for personal transportation. The Jet Pack speaks to individuality whilst the Driverless Cars speaks to collectivism – the individual subject to an algorithmic calculation!
The BBC delved into the subject – I thought it was an informative and sobering report.
Demographics are going to play a significant role in pushing adoption of this technology. There’s a whole lot of baby boomers who will increasingly have difficulty passing DMV vision tests in the near future.
When you’re faced with losing your drivers license because you can’t pass the vision screening, or you’re responsible for providing care for an elderly parent who can no longer drive, you’re going to go from skeptical to eager early adopter in very short order.
Personally, I’m rooting for the unicorns on this one.
Lordie. You must be under 30.
You clearly don’t know anyone with aging parents. First, there are things like glasses and eye operations like cornea implants that allow people with crappy native vision to see. The only way people really lose vision so that they can’t see with correction well enough to drive is macular degeneration (rarely gets bad enough to rule out driving; my father had it by virtue of commercial fishing for a while) or glaucoma (easily treated if caught in time). Your idea that a bunch of middle aged people can’t drive because their vision has gone kaput is laughable.
And the time aged parents they get to the point they can’t drive, which is usually in their late 80s due to slow reflexes and deteriorating coordination, they also should not be living by themselves. They will be in a retirement complex of some sort, like a facility that has independent living (apartments), assisted living, and nursing home care. Places like that have busses to take residents to events, often have ATMs and shopping on premises, take deliveries from pharmacies, and car service for residents to go to doctors and do other errands.
The last thing someone managing an aging parent would want is a driverless car to serve as an excuse for a feeble parent to continue to live alone. My mother is doing her own version of being in that category. She is driving me crazy because she can no longer get up off the ground without some picking her up (and I really do mean hauling her off the ground not just giving her a minor assist) and she has bad balance, so she could easily fall, break a bone or otherwise seriously hurt herself, and be unable to alert anyone that she needs an ambulance ASAP. Yet she refuses to move into a facility and instead gets by between having a home health care aide help her run errands once a week.
BTW one of her friends is 94 and he drives just fine.
When they can build an android, that can walk on city sidewalks, and get in line for a coffee at Mcdonalds , then they can build self driving cars. Until then.. pure and total nonsense.
I read an article that says that can’t run in the U.S.
Because our roads are so bad the cars don’t know what to do.
They might run fine over in China who has tens of thousands
of miles of new roads and highways but over here, LOL.
I just want to add a personal gripe- following the car in front of you, matching the speed.
THIS DRIVE ME INSANE.
I don’t want to drive someone else. I don’t want another car, ESPECIALLY at night, anywhere behind me. Get in front, or fade away.
The number one cause of being in an accident with another car is being near another car. Don’t be near another car if you don’t have to be.
I drive way too much, interstate to rural roads. Where you can be- stay alone, away from other cars. When you have to pass someone- DO IT, Quickly. Get away from them as soon as possible. I don’t know how many times I’ve had people sneak up in the left hand lane on not very busy interstates and PARK in my blindspot.
I blame Massachusetts. Is the m@$$h0!e word still banned here?