Top Myths of Robocars (and why V2V is not the answer)

Topic: 
Tags: 

There's been a lot of press on robocars in the last few months, and a lot of new writers expressing views. Reading this, I have encountered a recurring set of issues and concerns, so I've prepared an article outlining these top myths and explaining why they are not true.

Perhaps of strongest interest will be one of the most frequent statements -- that Vehicle to Vehicle (V2V) communication is important, or even essential, to the deployment of robocars. The current V2V (and Vehicle to Infrastructure) efforts, using the DSRC radio spec are quite extensive, and face many challenges, but to the surprise of many, this is largely orthogonal to the issues around robocars.

So please read The top 10 (or so) myths or robocars.

They are:

  • They won't be safe
  • The big issue is who will be liable in a crash
  • The cars will need special dedicated roads and lanes
  • This only works when all cars are robocars and human driving is banned
  • We need radio links between cars to make this work
  • We wont see self-driving cars for many decades
  • It is a long time before this will be legal
  • How will the police give a robocar a ticket?
  • People will never trust software to drive their car
  • They can't make an OS that doesn't crash, how can they make a safe car?
  • We need the car to be able to decide between hitting a schoolbus and going over a cliff
  • The cars will always go at the speed limit

You may note that this is not my first myths FAQ, as I also have Common objections to Robocars written when this site was built. Only one myth is clearly in both lists, a sign of how public opinion has been changing.

Comments

Brad -- Just the other day I saw a large graphic -- crammed with simple visualized statistics -- in favor of robocars, published by an insurance company, or cooperative group of them. In other words, contrary to any myths about insurance fears, it seems that insurance companies are in fact eagerly awaiting the day that we crazy mobile-phone-using, hamburger-eating, makeup-applying, text-message-sending, pop-song-singing, child-snack-administering idiots get our hands off the wheel altogether. They're basically telling us: "Y'all stink as drivers, and good riddance." I can't locate the graphic at this moment, and can't recall whether it was online or offline, but if I see it again I'll apprise you of it.

Some insurance companies are afraid, some are ready to face it. They know it is planned to reduce claims. Oddly, in California it is not allowed to reduce premiums very much but that could change with another ballot proposition (though nothing less than that.)

I enjoyed your point about moral decisions. Can you point me to some of the sources you used? Who's been making this argument, and what are "the numbers" you refer to? Thanks!

http://nextbigfuture.com/2013/01/robotics-cars-and-employment.html?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+blogspot%2Fadvancednano+%28nextbigfuture%29

This article is about how many people make a living from driving trucks. This estimate is low, since you also need to include the people who drives taxis and buses. We need to come up with a plan for all the people who will be become unemployed by robotic vehicles.

This isn't different than any other situation when technology makes some jobs obsolete. Where are all the lamplighters, now that we have electric lights? For a while, rules said that even electric trains had to have a fireman on board, though there was no coal to shovel.

Most jobs 100 years ago don't even exist today.

We are not talking about replacing normal cars with robocars overnight anyway.

Because we talking putting 5 million people out of work in the work in the United States alone. For people who didn't bother to do the Math. That is enough to cause a depression. So having a plan is an excellent idea.

It's true, some jobs will go away. It won't be all at once, but over the course of decades, a lot of jobs will be supplanted. This is not the first time that's happened, nor will it be the last. This does not mean it doesn't make sense to think about it, and plan for it at the policy level.

History has shown such technological job displacements are handled by society, though they may be much harder on any given individual. Driving is not a high training career for most -- for cab drivers it's not a career with much training at all. So we can't even say, "If you are planning for a career in driving consider something else" because I doubt many of the cab drivers planned for a career in driving.

However, all that said, what would you suggest we should do about it, other than make policy plans as I suggest above? Should we slow down the development of the technology to protect these jobs? In particular should we consider the million people around the world killed by human driving, and the hundreds of billions of hours spent driving by non-professional drivers who could be doing other things instead? From an economic standpoint, far more human time is spent driving by people just driving themselves than all the economic output of the professional drivers.

People bring this up, and it's not an unreasonable thing to bring up if you don't think people are already aware of it, but I would like for people to say something more concrete than "think about the jobs that will be lost." Folks are already thinking about it, and it's been listed on my downsides page for years.

I don't understand why you think it take decades. As near as I can tell a robotruck, robobus or a robotaxi will have a positive ROI of a few years. Why will any for-profit business buy a manually driven vehicle when a robotic version is available? The biggest resistance will be from private and public unions. Anyone that has to compete against a robovehicle will sweat bullets. A comparison situation is the IPAD. It is hard to remember that it has been on the market for less than three years. Laptop and desktop computers have lost a huge amount of market share.

The first market will be independent truckers. They can operate their trucks 24 hours a day without paying a second driver. The 1st generations robotrucks won't even have to operate on surface steets. Just operating on limited access highways will generate huge profits.

It will take a long time for all the jobs to vanish. In part because of resistance, and conservatism. Some things to iron out on delivery, too. It's rare for things to happen overnight. The taxi unions in Las Vegas have managed to keep a lot of things from competing, even hotel shuttles.

One issue with trucks is nobody wants to start first on big heavy 18 wheelers, the risks are higher.

Brad,

It's such a relief to read your comments about robocars. 2012 was a watershed year for news coverage. At the beginning, it was "wait'll you hear about this wacky idea!". At the end, it was just another news item. But a lot of the coverage is just wrongheaded, and it seems almost every article has errors. The V2V issue is pretty irksome, when V2V is neither necessary nor sufficient by itself. One author said the road trains were a more impressive achievement than the Google car; I don't understand how teaching a car to follow another one on a freeway is more impressive than teaching a car to drive by itself anywhere. People seem to trust and root for a system that seems mechanical rather than a car's intelligence.

But it's current planned uses involve a lot of effort for modest return. It would be nice to figure a way for it to evolve more naturally but at this rate, they will probably lose most or all of the spectrum allocated, making that harder.

Are V2V and driverless technology going to effectively solve 2 separate problems?
Reducing the 1.3 million deaths a year is likely to be achievable via automated driving.
V2V on the other hand is likely to be necessary for platooning which will increase road capacity and greatly reduce energy use. This assumes a reduced autonomy with a human piloted lead vehicle

The questions I would ask are:
Which of the goals are more important? Saving lives or resources?
Which technology can be brought to market readiness more quickly?
Would platooning save lives and how many?
If platooning with a human lead driver is simpler, is it easier to get from there to driverless vehicles than from our current position?

Not sure If I'm way off target, but fascinating blog, and I can't wait for driverless vehicles to become a reality.

The problem is V2V does very little. While robocars or self-driving cars (please not "driverless" -- there is a driver, it's just a computer) will put in V2V radios once they see a value for it, platooning is a tough technology to make work because you have to find your platoon.

To find a platoon you have to find a bunch of people all wanting to take the same trip, and a leader. If you arrange it in advance (like a carpool) you can set it up. It will be harder to find it ad-hoc on the highway.

Platooning provides a modest fuel saving (10-30%) but it has a number of issues, and the largest one is a safety risk -- if you have a failure, you might crash a whole platoon. So I expect that sort of thing to happen much later than robocars, not before.

Brad, I don't think governments can or should allow robocars to be programmed to contravene traffic laws, including speeding. Either the speed limits should be changed, or robocars be forced to observe them. It does indeed become a liability issue the first time a robocar set to speed by its passenger hits and kills a pedestrian (a 5 year old child, say). Who is liable for the manslaughter? I don't think it is viable for vehicle makers to allow passengers to induce the vehicle to contravene traffic laws...

Well, if the government is allowing it, it is not contravening the laws.

It is already the case that different classes of vehicles have different speed limits. Trucks, vehicles with trailers have different limits from cars. What matters is what speed is rated safe for that vehicle. If a robocar maker can demonstrate safe ability at 75mph -- better than humans do at 55mph, for example -- what precisely is the reason not to allow that?

There is a strong reason to allow it, however, and it's not so the cars can get places in a hurry. In fact, unmanned robocars generally are not in a hurry. The reason to allow it is that in many countries, including the USA, the human drivers routinely travel 10-15mph above the limit, though it varies by highway. Those humans take the risk of a ticket. The reality on many highways is that the risk is super low at 10mph over the limit, the police focus on more serious speeders, and so 99% of the cars on the road will often be going over the limit. And guess what: we, the police and society have decided that's what we want.

The problem is if you introduce a vehicle deliberately going slower than the other traffic because the written law says 55 and the unwritten law says 65, you actually create something that will impede traffic and possibly make the road less safe (because of the reactions of other drivers to it) rather than better.

I agree with you, better to have the written law and the unwritten law match. Problem is, if you raise the limit to the real speed, people are now conditioned with "the law deliberately understates the speed, go faster."

Better to do it like in France. There, the limit is 130kph and people do not exceed it much because tickets are issued. Or Germany, which has no set limit -- each car is responsible for choosing its own safe speed, and people never pass on the right and people pull over if they are going too slow for their lane -- and they have a lower fatality rate than the USA.

"Or Germany, which has no set limit — each car is responsible for choosing its own safe speed, and people never pass on the right and people pull over if they are going too slow for their lane — and they have a lower fatality rate than the USA."

The fatality rate is much lower. Germany has fewer fatalities than Texas, which has less than a quarter of the population in a bigger area. This is due to many factors: more training in drivers' ed courses, good public transportation so that not everyone has to drive, better cars etc.

There is no speed limit only on the Autobahn, not on other roads. Even there, about half of it has a speed limit because of hills, curves etc (so that one can't react quickly enough at high speed if something suddenly comes into view) or, in some cases to reduce noise near residential areas (sometimes only at certain times of day).

By the way, a speed limit on the Autobahn would reduce Germany's carbon footprint by about one half of one per cent. Banning all conventional light bulbs would have a greater effect. (About 1/3 of CO2 comes from traffic, about 1/5 of traffic (distance driven) is on the Autobahn, 1/3 of that is from heavy vehicles for which there is already a speed limit so that leaves about 4% coming from cars. But only about 1/2 of the Autobahn has no speed limit and even there the average speed is about the same as if 1/2 drive at 100 km/h or whatever and the rest at 160 km/h or more. So people driving fast are responsible for 1% of the CO2. If a speed limit cut their emissions in half, that would thus reduce Germany's footprint by half. Many, many other things would have a greater effect. Banning SUVs would be at the top of my list.)

But the point is, the speed limit isn't some magic number. It is different for different roads (which road engineers do follow but politics also affects what limits they set) and different conditions, and different types of vehicles and different drivers. All have a different maximum speed at which they reach acceptable safety. The Autobahn is just an example of that. My personal view is that the safest thing to do is for the robocars to go at the speed of traffic in their lane, so long as they can handle that speed safely.

http://www.actuarialoutpost.com/actuarial_discussion_forum/showthread.php?t=254628&page=19

These guys think that making a car that drives itself is the easy part, and that adapting to it is the real work.

He calls most of them not myths or straw men. But the reality is that I read new articles every week about robocars, and these myths appear in almost all of them. So I wrote the article to address them.

The ones the poster disagrees with -- well, I guess we disagree. The disagreement is mostly about whether they can be made safe or how hard it is, and whether people will trust it.

Well, it's absolutely true that it's uncertain if they can be made safe and when. That's not the myth. The myths revolve around declarations that doing it is close to impossible for this or that reason. This declaration is the myth.

As to whether people will trust it, I know they will over-trust it both from studies that have been done on this and on actually riding in cars with people. So if somebody has some studies that show the opposite, I would be interested.

Brad,

I can't wait for auto-driving cars, but in the mean time, is there room to think about "zero-visibility driving" first? Is there an industry term and/or an ongoing effort for this type of thing already?

You probably already get it, but just to be clear: I'm talking about a video-game display, showing other vehicles around me, plus the current speed limit (including road conditions and school-zone schedules), plus nearby road signs, plus the state of the traffic signal directly in front of me, and so on. This doesn't seem a technical stretch (tho we might need to install transmitters at every traffic light). I've seen some half-hearted efforts to compile a speed-limit database out there, but I haven't seen any complete products, open-source or otherwise. It seems like a no-brainer to me.

Not to beat a dead horse, but with something like this I could choose the lane with fewest cars at the upcoming light, I could schedule a lane-shift to coincide with the right-turning car next to me (assuming V2V), or I could scoot forward if I knew that I was blocking ten cars behind me (e.g., because I'm turning left and I'm not sufficiently pulled forward into the left-turn lane). And so on.

Is this type of thing already out there, in whole or in part?

Well, there have been experiments -- but just experiments -- on driving for the blind, but the reality is that the blind will be better served by full robocars.

Zero visibility (like fog) blocks the LIDAR as much as human vision, and radar just isn't enough to drive like that on. V2V is not suitable for this either, as it will never have 100.00% penetration.

There are radars out there that can produce a more LIDAR like view, but it needs a lot more bandwidth than is allowed in the spectrum allocations today for radar.

Add new comment