The Driverless Car Revolution Should Not Begin with Cars
perspectivezoom.com> Compared to driverless cars, driverless trucking is technically way simpler. When going from one warehouse in a sparsely populated area to a another warehouse in a similarly sparsly populated area, there’s a lot less that can go wrong.
Not quite. Long haul trucking has far more things to go wrong than a short jaunt from two packed urban spots to another. A lot goes wrong on these trips. Flat tires, accidents, bad weather and - of course - other drivers.
Case in point: a long haul truck from Oakland to Chicago has to tackle Donner Pass, the Wasatch, Sherman hill (the Wyoming continental divide) and then finally gets a flat stretch across the Great Plains.
I would counter that the best start is short-haul intermodal: trucks that pick up a container and haul it to an intermodal rail yard, and so on. Those trucks have to deal with a short route and much fewer potential problems.
>the best start is short-haul intermodal
bingo. In container ports, long-distance drivers will generally dump their trailer to be shunted and loaded internally, and pick up another one. Separate cabs are attached to move things around within the port.
These could relatively easily become driverless. It's a highly controlled and automated environment. Every moment a ship is in the port costs money, so there's a big reward for improving reliability and shaving seconds off (un)load times. It might make financial sense even in Asian ports with low wage costs for drivers.
Same with mining or heavy industry. Where companies might now build their own railways, in the future they could build a road and run driverless trucks along it. Being outside of public road systems makes the legal situation easier, and you can always have a human jump in to drive the last mile.
No "in the future" about it... we're already seeing automated ore haulers in Australia's pit mines, for example. It's a perfect storm of a controlled, sparsely populated and dangerous environment, relatively low cost (thanks to fewer, larger vehicles, and automation costs rising slower than size), etc.
You can also have a convoy of 1 manned truck with n driverless truck that play follow the leader. You end up with a train model on the road infrastructure.
You can already do that with a road train.
I suspect a mile of highway has fewer hazards than a mile of Manhattan.
Trucking from the suburbs of a port city to the suburbs of another city could be 98% highway. Not many toddlers will be walking into traffic they way they might on a residential street.
>Long haul trucking has far more things to go wrong than a short jaunt from two packed urban spots to another.
Some road side service response personnel could alleviate that. I personally think automating long haul would be a great thing, I've seen a lot of hair raising driving by truckers on I-84 and if we can't expand rail transport or zone commercial/industrial near rail access then I see this as the next best thing. Of course truckers would be quite vocal about any sort of change...
> Compared to driverless cars, driverless trucking is technically way simpler
This person has never driven a truck or a bus[1]. The amount of stuff that can go wrong in a truck or bus, plus the amount of damage an error would cause is well above a car.
I think the whole driverless car should have started with smaller-than-car vehicles. I wish they had started with simple package delivery vehicles[2] and learned from them.
1) I did have a bus license in my poorly spent youth
2) something like http://badgerlandminitrucks.com/specs.htm but smaller. Vehicle drives to A, texts, accepts cargo and payment, leaves for B, texts person at B, B verifies ID, door opens, vehicles goes to next site - probably work OK in a city
The only flaw I see with beginning with semi-trucks is the perception of driverless cars to the masses and the potential damage a driverless semi-truck accident could incur vs a driverless Prius or something smaller. Not saying there won't be a market for it in the trucking business, but seems smart to start on the smaller cars just for the sake of perceived safety.
Exactly what I thought. There is an awful lot of damage a semi-truck can do if things go wrong. But, then I realized, that there is not necessarily a need for the "trucks" to be big. Maybe you could go the Roomba route, and make them much smaller than traditional trucks, maybe even smaller than conventional cars. Since no driver is needed, you can scale out with many of them, rather than having one very large truck. Maybe that approach does not work for everything, but somehow I think having a large bunch of cargo ants is much less risky than a single very large cargo elephant.
If the trucks are small, they're no longer compatible with standard-sized shipping containers, which would throw up a significant barrier to adoption.
There's a lot of cost in the mechanics of the truck and the fuel which make full-sized trucks more economical. Not to mention the convenience of not having to load and unload lots of small deliveries. I would be very surprised if self-driving mini-trucks were anywhere near economically competitive with human-driven semis.
Any time I read these utopian pieces about the driverless car nirvana that is around the corner I roll my eyes. If there's anything we should learn from history it's that nothing is as simple as it seems, especially if it's something we want.
There will be unforeseen technical issues that are not apparent with a single test car or even small numbers of test cars. There will be accidents, and we'll have lawsuits against Google or other manufacturers. Those will have to play out, so we figure out who is really liable when a driverless car crashes. Trial lawyers will flock to these like sharks to a bleeding fish.
We may get to a point where driverless cars are safer than human-operated cars, and are accepted by the public. I think it's at least 50 years away.
It's not 50 years away, it's closer to 5 years away. In 15 years they will dominate the roads.
Google's cars have already demonstrated that they are safer than human drivers with 99% statistical certainty, they have over 750,000 miles of accident free testing on public roads under their belt.
Because Self Driving Cars stand to reduce traffic accidents by 90% or more, it is likely that they will be afforded certain legal protections against the sorts of class action lawsuits you're suggesting, much as they do for vaccine makers, under the auspices that in spite of the risks involved for manufacturers the net benefit to society is too great.
> Google's cars have already demonstrated that they are safer than human drivers with 99% statistical certainty, they have over 750,000 miles of accident free testing on public roads under their belt.
Has Google driven their car in snow storm[1] or with snow/ice on the ground?
1) I'm talking basic heavy snow, not a blizzard
They have not done serious work yet with driving in snow and other inclement weather conditions, this is the phase they're entering into as we speak. They've suggested their cars will be ready for the public in 2017, I guess we'll see if they meet this deadline or not.
So they haven't driven in road conditions that cause quite a bit of accidents but you state they "demonstrated that they are safer than human drivers with 99% statistical certainty, they have over 750,000 miles of accident free testing on public roads under their belt."
Is the success you quoted adjusted for lack of adverse weather conditions and is it published?
Agreed. The corner cases are very difficult.
Rio Tinto Deploys Driverless Trucks http://www.driverlesscarhq.com/rio-tinto-deploys-driverless-...
Rio Tinto, you mean.
yeah, oops.
I think what will come first is driverless subway cars. They've already had this in some places (https://en.wikipedia.org/wiki/List_of_driverless_trains), but they're just replacing a regularly-scheduled human-driven train with one without a driver.
But think about what can happen if you have a system designed to not require a driver at each subway car. You can have smaller cars, called on demand. You don't need to wait for the next car to come. This is especially a benefit during late night hours, when you might have a subway come only every 20 or 30 minutes. This is made worse if you need to transfer.
Once you get cars running along existing lines with small cars, you can eliminate subway lines entirely. Just hop in a subway car, tell it where to go, and it gets you there. No transfers, no thinking of the best lines to take, just get in a car and it'll drop you off at your end location. It would eliminate taxis for many people, which would be one of the reasons this plan would create controversy. Also controversial is elimination of conducting jobs.
I see a few reasons replacing subways will work better than trying to replace cars:
* One organization already owns all the infrastructure for a subway: they can just choose to replace it, and get the laws written for themselves.
* Rails mean you don't have to worry as much about steering, or cars next to you turning into you.
There are already many driverless subways in existence (especially at airports). Many driven subways (BART in SF) are actually driverless, with a driver there just to close the doors and in case of emergencies.
http://en.wikipedia.org/wiki/List_of_driverless_trains#Unatt...
Good luck implementing totally driverless subways on currently driven subways, though. Unions will totally block the necessary incremental steps, so it would have to be an all-or-nothing back room skunkworks project from a public transit authority.
Fatal flaw in this plan: unions. New York has already tried to switch to driverless subways, but progress is slow due to constant union interference.
I totally agree that driverless trucks are the future, but I don't think it will come easy. The International Brotherhood of Teamsters is one of the largest unions in North America, and one of the most politically powerful. I think that they will fight this with everything they have to keep humans getting paychecks.
Exactly. One of the main barriers to the adoption of driverless cars is going to be politics. Starting off by threatening a bunch of unionized jobs would be political suicide. I can see the political cartoons now: Terminators with glowing red eyes driving trucks, running down crowds of helpless union members.
The number of longshoremen needed to run a modern port has decreased by a factor of 10 over the last 30 or 40 years. Because of their strong unions, the one's who remain get paid very well. Unions don't really have a strong bargaining chip once their labour is no longer needed.
yet, it already exists.
http://www.mineweb.com/mineweb/content/en/mineweb-fast-news?...
i hope the shock of all taxi/trucking/pizza delivery/etc jobs being eliminated leads to a substantial rethinking of employment in general
Except the power of going on strike has been completely disintegrated with an automated fleet of trucks sitting back ready to be deployed.
Disruption will come, and they won't be able to stop it.
He's not referring to a strike, but to politics.
Because strikes have never been used as political tools by organized labor groups…
All cats are mammals but not all mammals are cats.
Nothing like a persuasive comment to save the increasing obsolescence of human labor applied in the aforementioned way from the automated chopping block…
Driverless trucks is a real opportunity that could save a lot of money in shipping.
But it could be exploited. If a person walks out in front of a truck, the automated truck better stop. Which an unguarded truck would become easy to rob with little risk. A flat tire would mean the truck calls in for help and is stranded defenseless on the side of the road. That said a fair amount of loss could probably be absorbed with the savings.
I expect long haul truck drivers instead to turn into truck captains. They're present but not for the driving. The truck can drive through the night while they sleep. The captain will handle weigh-ins, emergencies and other road tasks.
Honestly, I would rather this situation occur to a "defenseless" driverless truck than to one with a driver who might be injured or killed in the process. I highly doubt the rate of people stepping out in front of semis to rob them would increase drastically in any case.
whereas a manned truck in the same situation would just run over the guy? And what about locks, cameras, burglar alarms etc?
Surely a manned truck of today is easier to rob. You don't even need to move the goods, just steal the whole truck and be done with it.
by the time you're crazy enough to step out in front of any sort of semi in hopes it will stop, so you can rob it, i don't think you're doing risk calculations in a rational way.
Okay so put a mannequin with enough of a heater inside of it to pass for human on IR and strap it atop a roomba and then have THAT "step out" in front of a truck. It's a $1000 dummy and as long as you rob every 5th truck successfully (20%, but really you'll probably get 90% of them) you can make back all the money on busted dummies and more.
One human 'captain' for a convoy might make sense.
Since 2010, I keep saying that Singapore (nation state, in this case) will be the first "city" (with at least one million people) to adopt driverless cars in a short span of time. I would even accept bets :)
Municipalities with a lot of freedom will follow suit. I don't know if american cities will have an easy time fighting with unions, conservatives, taxi drivers, real estate owners.
Why real estate owners? Because once transportation gets optimized, "hot" areas of the city will cool down in prices, compared to other parts of the city.
Why does it have to be a driverless car revolution? It seems more likely we'll get there incrementally, adding pieces of the experience one by one as safety features. Eventually, with all the safety features enabled, the car will be able to drive itself.
There was a post about this a couple weeks ago about why incremental change might actually be less safe than a "revolution". Basically, if a car can handle 90% of the scenarios it faces, the human driver would zone out and have difficulty responding to the remaining 10%, especially if there's little to no warning of when that 10% will happen.
Yes. Also, when a company knows it needs to design a 100% driverless car, safety will necessarily be the primary concern when designing the firmware. While that's theoretically the case regardless, recent Toyota fallout[1] shows that car companies perhaps aren't paying enough attention to this given that they have a human driver ultimately making the decisions.
A problem with trucks as the leading-edge is that they're already a bit alien to the public and policymakers: resented in traffic, feared out of proportion to actual risks, underappreciated as some narrow corporate/union interest. If they're the test case, enthusiasm will be lower, regulation more costly/cautious, and the public vaguely suspicious, primed to assume the worst when anything goes wrong.
Plus, people drive like idiots around them. It is amazing what normally sane people will do in the presence of a truck.
Yep. People who know they're around a self-driving car may act weird, too: gawking at or challenging the software. (The young and reckless might even play chicken with autocars for thrills.)
> There are significantly fewer one way streets, fewer bicyclists, fewer pedestrians. The directions will be simpler on average.
While that may indeed be true on average, the real challenge in developing production-level driverless tech is properly handling the edge cases. Just because a vehicle may not often have to deal with bicyclists doesn't mean it never has to deal with them.
The major benefit of self-driving isn't saving the costs of a driver's salary, it's reducing accidents. Commercial truck drivers are skilled and alert, so it's hard to be safer than them. Average commuters are distracted and accident-prone, so you can reduce accidents with near-future technology.
This was just in the news today:
"Looking at photos of women on Facebook is what distracted the driver of an empty fuel tanker when he crashed into two fire department trucks and three police cars on Interstate 8 killing a public safety officer in May..."
http://lasvegas.cbslocal.com/2013/11/03/on-facebook-looking-...
And imagine how many could have died if it was a fully loaded fuel tanker.
I don't know which commercial truck drivers you've observed, but anecdotally, I've seen some pretty distracted / inattentive behavior from truck drivers myself.
Are you suggesting that truck drivers are somehow ubermensch who can stare at highway for 16 hours without getting drowsy or losing focus? Automated trucks can drive with perfect attention 24/7.
That being said, I disagree with the article -- trucks take longer to stop, are harder to navigate in constrained spaces, and have higher stakes than personal cars. I wonder how difficult it would be to rob automated trucks by simply slowing down to a stop in front of them, for example.
Do you have a source which shows that commercial truck drivers are less accident-prone, and by how much?
Not the OP, but http://www.fmcsa.dot.gov/facts-research/LTCO2010/2010LargeTr... - Figure 2 and 3
Thanks!
Alternatively, instead of trying for heavier vehicles with longer hauls, what about something like citywide pizza/package delivery? You've no need for passenger safety, so the size and weight of the vehicle can be brought way down. Speed could also remain relatively low, mitigated by the fact that deliveries could be dispatched immediately rather than aggregating trips. Not that a person getting hit by a 400lb vehicle at 20mph is a laughing experience, but accident damages would be mitigated and an emergency stop-everything would actually be feasible.
I was in the bay area recently and got stuck on 101 driving from Google to GoodData in downtown SF....
It made me think that the place to start is the car pool lanes on 101.
Just close them off to humans. Instead, make them available to driverless cars with one addition -- a driverless on-top tow truck.
You drive your existing car on it. Type a destination in on your phone and press GO. It dumps you an one of several exit points.
Ideally, you'd do this with rail instead of roads but that changes it from a tech problem (mostly) for an infrastructure problem (mostly).
As much as I'd like to read more of this article, the choice of font makes it pretty unreadable.
has anyone looked into what effect the driveless car revolution will have on the labor market?
i'm not a luddite. i'm just curious how many people will be affected professionally.
I giggled a little bit too much on this pun
[4] "Who will drive adoption?"