The approach that Starsky (http://starsky.io/) is taking seems to be a promising one. They employ truck drivers as remote pilots for the "first and last mile", i.e. the local roads and loading dock work that autonomous vehicles struggle with most. Other advantages include lower fuel costs (driving slower because AVs aren't subject to operator-hour limits) and probably also lower labor costs (you don't need to pay truckers as much if the job is safe and nearby to home).
If the Starsky model ends up dominating, this could all add up to a case where the price of truck transportation drops so much that demand for truck drivers actually increases, possibly even counterbalancing the pressure of individual truckers being able to manage multiple vehicles at once. This would be an example of the "Jevons Effect" https://en.wikipedia.org/wiki/Jevons_paradox
The long haul stuff should go on rails anyway. It's very wasteful to run it on the interstates. The majority of the fatigue damage to the highways comes from trucks, then there's all the rubber dust, fuel inefficiency, etc.
The trailers should just be picked up and dropped on flatcars, and trucks used for intracity last mile stuff.
I hear this all the time, and honestly in many cases a truck is the transportation of last resort. If you need 5 shipping containers of goods going from A->B all arriving at Time X; it's probably going by train.
An issue with rail (as pointed out by others) is how to haul the freight from the rail head (rail yard) to the end warehouses, and the extra time it takes to have goods go on two trucks and one train instead of just one truck. The other issue is that in the US it's really hard to hire and retain long haul truck drivers (it turns out most people want to go home each night after a hard day’s work).
Competition to hire drivers is fierce, and driver turnover can be 100%/yr...which means sometimes your drivers don't show up to pull your freight. Rail is also generally cheaper so if you can put your goods on the train you do.
As far as wear and tear on the highway: that's what weigh stations are for. They not only assign usage taxes, they also make sure weight is evenly distributed across the axles to limit damage to the road.
>As far as wear and tear on the highway: that's what weigh stations are for.
I believe WalterBright meant that even at the legal weight limit, trucks cause 10,000x as much damage as cars. In 2005 this amounting to a $60b taxpayer subsidy.[1]
This isn't to criticize Starsky Robotics (this inefficiency predates it), which is just operating in the environment it finds itself. And in general, inefficiency begets opportunity.
Kinda. Has more to do with axle load than total weight.
You could also say that the research is pretty incomplete. All the known studies have been performed at highway speeds...it's entirely plausible that speed exaggerates the damage disparity.
Right or wrong, I don't think you have to defend that point. The government sets these parameters under which you (or truckers) do business. It's not your responsibility to maintain the highway as long as each truck complies with regulations. And it's not your fault.
American taxpayers subsidise many other things that don't quite make sense (to me), like corn production. You don't see people blame farmers for doing what's best for their farm and family.
Hey are you all trying to sell your hardware and software or are you trying to create a whole shipping company? Who are your competitors? I've been trying to do some research on companies in this space but it's been a bit hard to find info on anyone except Otto. If you've got the time I'd like to ask you a few questions, my email is in my profile.
This comment invariably pops up in these threads; If it was going to, it already would have.. It's too late now.
Would it have been a better solution? Probably, but the investment required (laying down the rails, land buyouts, copious use of emanate domain, political greasing of the wheels, etc...) has always been prohibitive. The gains would be shared between many disparate groups that would need to fund this together, taking years to decades for any ROI - certainly a longer time-frame than most publicly traded, quarterly measured, corporations can afford. At the least this would have required some government subsidization.
With driver-less tech around the corner, returns on this investment for any one company become even more dubious, and thus make it less likely to happen. I think the parent comment here will always get some up-votes, but that is as far as I see this going anymore. Maybe it'd go farther if we threw in words like 'Hyperloop' ;)
The rails are already there. When I drive down I5, you can see them.
The reason this sorry state of affairs exists is because truck traffic is heavily subsidized (highways paid for with taxes) whereas rail traffic is heavily taxed (and the rail companies have to pay for the tracks).
It could be improved by increasing weight taxes on trucks, and using the proceeds to subsidize intercity tracks.
At least in the United States, this is incorrect. Long haul trucking is expensive compared to shipping things via the rails. The reason things end up being shipped via truck is that its faster, plain and simple. Shipping via rail car is slow and has no guaranteed arrival date. Or should I say, dependable arrival date. And from a logistics stand point you have to engage a last mile trucker at the rail head nearest your true destination to haul your item to the destination. Since you're unlikely to be shipping an entire container/rail car, you'll need to pay extra since the delivery driver will have to drive to your destination then 'dead head' (i.e. return without a load) to the rail yard.
You're arguing past parent. He stated not that trucking is cheaper, but that it is heavily subsidized. In almost all cases, rail would cheaper, and some things will take advantage of that. Likewise, in all situations trucking will be more expensive,yet there are use cases where the reliability and speed are justified (just like couriers are sometimes justified.) The difference made by the subsidies is all in shifting the margin.
Things that would be toss-ups or wins for rail are instead pushed to trucking. This would just be wasteful, and not so pernicious, were it not for the serious negative impacts on the layout of our towns that result from favoring trucks so heavily.
These are just organizational problems that could be worked out if the financial incentives were there, instead of tilted heavily in favor of trucking.
Prior to the rise of the Interstate Highway system, most warehouses were on rail lines, or spurs. You'll still find the trackage, or at least the traces of them, in many urban areas throughout the U.S.
In San Francisco, the region near Potrero Hill by the old Fashion Design Centre. The odd street patterns and building perimeters show the former track rights of way.
Even in France, with paying interstate (called autoroute here) and subsidized rail (it was a state monopole until recently), most of transport of goods is by truck.
I think it's mostly a last mile problem (you'll need big trucks to carry from the station), compounded by the fact that most of the train station of cities and I don't think they is a lot of dedicated cargo train station. Also you do not control part of the schedule: you are dependent on the departure/arrival time of trains. And don't forget the strikes...
Europe generally prioritises passenger transport over freight transport, but there are rail freight terminals -- probably not anywhere you're likely to visit (example [1])
Rail traffic is allocated on a timetable, for both passenger and goods trains. The time and route is called a "path" in British English, and reserving one costs money.
A power station might have a daily path from a port to their power plant, as they can reliably use all the coal.
Rail freight companies book many paths between freight terminals, container ports and so on and mix together goods from many customers to run on trains on these paths. Therefore, the timings are reliable, potentially more reliable than by road, but the flexibility is less.
If there aren't any goods to move, the company doesn't need to use the path. (I used to live close to a railway line used to transfer trains containing nuclear waste. The reprocessing plant had a path to run a train every day, but only used it every couple of weeks -- presumably, they don't like leaving the waste at the power station any longer than necessary.)
I believe SNCF was also know for loosing freight a few years ago. Might have improved these days since I assume most cars must have some sort of GPS. Timing was another problem since freight had to yield priority to passenger traffic, which is pretty dominant in France.
You also have cases like Portland where they shut down the shipping terminals which mean that all the things that would move by rail instead moves by truck now.
The amount of semi traffic from Portland to Seattle on I-5 has exploded, it certainly made driving that stretch a lot less enjoyable.
There's a lot to this. But it's not just about lobbying, it's also about unions and votes. The American auto dealerships likewise have a lot of what's described as "lobbying power" when really it's more that they have a lot of people that want to see politicians protect them.
Many trucks in the US have a bumper sticker that says, "Without trucks, America stops." Teamster strikes, like the León Vilarín teamster strike in October 1972 in Chile, are a disaster for any country and create national emergencies.
The current rail network already grows and shrinks with demand so long term you it can continue to expand without massive investments.
IMO, Automated trucking makes trains more not less useful. The core problem with using trains is speed plus how much time and effort loading and unloading takes. With automated trucks you can significantly reduce this friction.
What makes you think this isn’t already happening?
Norfolk southern runs behind my house and I see semi trailers on top of rail cars all the time. I see UPS trailers on trains almost every time I watch one.
Refrigerated semi trailers are common on trains too. I’m guessing they have food in them.
There's very real advantages to long-haul trucking over rail. Shipping time costs logistics operations real money in terms of inventory costs, flexibility, and responsiveness to market conditions. Cargo ships are even cheaper than rail, but shipping costs aren't always the dominant concern. Similarly, some goods get air shipped even with the higher cost of doing so.
- For one, you're talking about human drivers covering first and last mile, which is a tiny, tiny fraction of the route that a truck covers which is why the Starsky model works well, in theory anyway. That means one given driver can do the work of probably a few hundred trucks in their last mile phase, even if they end up with some downtime in the middle, it's not even close to balancing. You're still talking about a possibly slightly larger market being overseen by a much smaller workforce.
- Secondly, this completely ignores what could be the silver medal effect of truck driving becoming automated; the death of millions of small and large businesses that operate basing heavily on the incomes of truckers, everything from truck stops and diners, all the way to motels, specialty truck accessory companies, on and on and on. These are the ripple effects that many people aren't looking at when they're thinking about this, it isn't just trucking, it's one of the last jobs available to you with naught but high school under your belt that earns around $40k. And that's not even going into how many businesses do a ton of business with truckers that earn that wage, which is usually much higher than their neighbors are earning.
- Thirdly, if all trucking is done from various Starsky-like operations, then truckers in general will be moving into more consolidated living conditions. Again trucking is one of the VERY rare jobs that requires minimal training, pays a comfortable wage, and is available pretty much anywhere in the States where there are roads and people who buy products. If you want to work as a software developer, chances are you'll have to move to Silicon Valley, San Francisco, Austin, etc. but if you're looking to be a trucker, you can go pretty much anywhere and have fairly solid employment opportunities. For now.
If you want to work as a software developer, chances are you'll have to move to Silicon Valley, San Francisco, Austin, etc.
Not that I disagree with your basic point, but that's a pretty narrow view of the software world. There are jobs for developers all over the place, even in podunk towns with a few thousand residents. Granted, that town might have like 3 total jobs for programmers, but a. that's an extreme example and b. "long tail". Consider all the mid-sized cities out there, as well as "non tech hub" big cities and there's plenty of places for programmers to live and work.
> Thirdly, if all trucking is done from various Starsky-like operations, then truckers in general will be moving into more consolidated living conditions.
why? if the controls are network traffic, and one driver being right next to another driver doesn't add value, why not spread them across the country? it'll reduce control latency, increase reliability (can't have all deliveries stopping if the fiber to one building is cut), and make it easier to have drivers awake when the deliveries are tending to take place (certainly some locations might take deliveries in the middle of the night, but mostly they're going to happen when businesses and such have employees available to receive the materials.)
That's almost certainly a temporary effect though - if the trucks are being outfitted with the hardware necesary for full automation, full automation is only a software update away and the pilots will only have a job for the relatively short period where automation isn't good enough to handle that portion of the journey.
> this could all add up to a case where the price of truck transportation drops so much that demand for truck drivers actually increases
The problem is that local drivers average about half the pay ($30-40k) vs long-haul drivers ($70-80k). Even if there is more demand for local drivers it likely won't offset the wages lost.
There are a lot of unknowns here. The factors putting downward pressure on trucker salaries are:
- Remote operation of trucks is an easier lifestyle, so more people would be willing to take the job, driving down salaries
- Fewer human truckers are needed per truck on the road, so for a given level of trucking activity, fewer jobs exist
Some potential countervailing factors:
- Price of trucking plummets, potentially jacking up demand (Jevons Paradox) depending on price elasticity of truck transportation
- Managing multiple vehicles may end up being more cognitively challenging than driving a single truck, reducing the pool of available workers and driving up salaries
The values of these parameters (and how they change over time) will be crucial to determining the net impact to trucker salaries and employment numbers. My intuition is that salaries will go down dramatically, employment numbers will go down modestly, and overall trucking activity will expand.
> Managing multiple vehicles may end up being more cognitively challenging than driving a single truck, reducing the pool of available workers and driving up salaries
Tangential to this point: If this mode of truck operation becomes prevalent, I could potentially see a path where warehouses/loading docks get employees certified (a la forklift truck certifications) to operate the truck on premises to handle the "last mile" (i.e. on-site) operations. Now you've completely decimated the number of trucker jobs.
I think you're drastically underestimating the difficulty involved in manuevering an eighteen wheeler around the premises of a typical business. There's no way it would be cost effective to train people for that so that they can spend most of their time doing something else.
Typically the gains would be shared through lower prices to customers and higher profits. Lower costs might also increase demand for services and goods, growing the industry and increasing employment elsewhere in companies, though that latter is likely to be a relatively small effect.
That's right, it frees up $1bn for other uses. This is why automation, ever since the industrial revolution, hasn't lead to persistent mass unemployment. The capital freed up by more efficient industry was used to create new industries and jobs that didn't even exist before. There have been some painful adjustments, but the more flexible an economy is the more rapidly it adjusts. Historically this has been a US strength.
Rich people spend much less of their money on stuff than poor people do. Another thousand to a millionaire goes on the pile. Another thousand to a single mom goes on kids' shoes and car repairs.
People who have no hope of ever retiring from their low wage job that they live in fear of losing to automation might count anyone with a pension plan/retirement fund access (and even insurance) as rich.
> but taxing gains in productivity is a perverse incentive.
Not at all. The end goal should not be productivity onto itself. If those gains are being scooped up by a sliver of the population, we change the rules to benefit all. One person, one vote.
As automation automates away more jobs, the cohort of people without jobs grows, forcing regulations (through democracy) to change to move towards a citizen's dividend funded by a tax on automation. The system will reach equilibrium, or there will be a revolution.
If you don't want your automation lifting the tide of all boats, someone else will come along and build it.
> They employ truck drivers as remote pilots for the "first and last mile"
It would make even more sense to employ a "local" pilot with local knowledge for the first and last miles, kind of how tug-boats help large container ships in and out of ports.
If Uber is right about it's autonomous taxis, and Tesla's cars really will be able to come when you whistle, then last mile driving and docking are surely solvable problems.
I'm generally an optimistic person, but I think autonomous driving is going to seriously disrupt our societies. I simply cannot see where the huge workforce of unskilled males driving trucks and taxis is going to be soaked up. The industrial revolution did it last time, but anything you can suggest as a potential equivalent (colonising Mars?) will be demanding smart flexible people - and an awful lot of automated systems.
I'm picking the mass unemployment will lead to civil war in the US (where all those unskilled males have semi automatic guns) and universal basic incomes in more civilised countries.
Come to think of it, this would make a good comic or sci-fi film.
A mob of former truckers, farmers and Uber drivers armed with bump-stocked AR-15's wearing Make America Analog Again hats and fueled on pillaged Soylent trying to overtake Silicon Valley and having to fight through an army of Teslas, Otto, Boston Dynamics, Weaponized Amazon Drones, etc.
Parallels also to the rise in the demand for horses after the introduction of the railroad.
The availability of long-haul transport increased transportation volume, but the "last mile" connection from origin/destination to the railroad depot couldn't be accomodated as readily. So: horses and carts.
Variously: Robert J. Gordon, The Rise and Fall of American Growth, Vaclav Smil, Energy and World History -- of which there's a heavily revised edition out this year, Energy and Civilization.
I really don't see why autonomous vehicles couldn't handle the first/last miles soon after mastering the open road. I think only regulation would made this a requirement.
For trucks that don't follow the same routes every day (e.g. FedEx/UPS/etc.), even humans frequently have a hard time getting it right. "Oh, no, it's the other side of that bridge that's only for light vehicles. Yes, you'll have to drive around for two miles in heavy traffic to get here. No, it's the entrance on the back facing the tall green building." Etc...
Until AI for vehicles massively increases, humans are going to be better at navigating the truck through smaller roads, abiding by weight restrictions that may not already be known about, having local knowledge of the area that AI might not have. Every delivery location is going to be a little different so you would need a really general AI to understand how to just get the truck up to the loading dock (if you are that lucky).
The long haul portion of the journey is much more suited to the AI of today, cars already have enough tech to get this done with pretty good results.
> Every delivery location is going to be a little different so you would need a really general AI to understand how to just get the truck up to the loading dock (if you are that lucky).
or, as someone with a loading dock, i'd be willing to pay to have someone come out with a bunch of lidar gear and whatever to model up my dock and surrounding area, so i could just provide a big data blob about my dock to the shipping companies, and have their AI trucks pull in correctly every time.
as opposed to worrying about the days when some new guy (with his local knowledge and general intelligence) shows up to deliver and comes in the wrong way and gets stuck for 2 hours.
(there's a new job for you right there, folks, since the model will probably need to be updated every year or two as the shipping companies change their AIs, and changes are made to the surrounding buildings.)
I expect many regular delivery end points will get set up with autonomous vehicle standardised parking bays, with programmed routes in and out added to the AI mapping databases. If you don’t have one of those, you’ll need to pay a premium for driver operated delivery or remote driver guidance.
The savings likely aren't worth it. Trains can pull a lot of cargo compared to a truck, so the marginal cost for a driver is a substantially smaller percentage.
I don't think it has anything to do with cashier wages, and more to do with customer perception.
Kiosks are already cheaper than people.
With minimum wage at $7.25/hour, and a Kiosk being able to replace 2+ shifts (16 hour days 7 days a week), that's $30K/year in savings, which would more than pay for a kiosk and maintenance.
Developing the software to run them isn't free, but spreading $15M in development costs among 14,000 restaurants is only $1K each.
But many customers would balk at having to order at a kiosk, so McDonalds will still hire humans.
> But many customers would balk at having to order at a kiosk, so McDonalds will still hire humans.
A lot of locations around me have both kiosks and staffed registers.
I used the kiosk once and thought it was annoying to work through. If you have a simple order like "A medium #3", then its an order of magnitude faster to order through a person. On the other hand, if you like to customize your burger, or order a several things a la carte, the kiosk might be better depending on the competence of the human at the cash register.
(Of course, whether or not your customized burger or list of a la carte items gets made correctly by the kitchen staff is an entirely different conversation...)
> If you have a simple order like "A medium #3", then its an order of magnitude faster to order through a person.
This is a user interface [fixable] problem, not some inherent defect in kiosks. You should be able to walk up to the kiosk, tap #3, wave your NFC payment at it, and walk away with a hamburger. Today's "Tap. Tap. Scroll, Scroll. Tap. Back. Tap. Tap checkout. Tap Ok. Insert credit card in this other machine. Tap confirm. Sign. Remove card. Tap done." mess is merely incompetent product design.
Also usability. Ever watched an older person try to get a drink out of one of those touch-screen drink dispensers with 100 choices of soda+flavors?
Same would happen with a kiosk order taker. Two humans talking can still communicate much more reliably than one human navigating a somewhat unfamiliar UI.
There's an initiative to move to autopilot for trains, or Positive Train Control. The US Gov't started the initiative in 2008 after a particularly gnarly train accident which killed 25 people. It was due to be rolled out in 2015 and we're still waiting...
For long-haul rail freight the economics don't matter as much as they do for trucks. 1-2 people can control a 7 mile train as opposed to trucks which have 1 driver per 50' trailer.
They exist for local rail and subways, and with the unification of the European signaling and braking systems, soon they’ll also dominate long-range rail
The most likely thing (imo) that will happen will be highway trains...there will be a real human driver in the front of the caravan, maybe just in a van or car, and then 1-10 self driving heavy duty trucks behind them. They'll draft off of each other and do 99% of the long haul route in an automated way.
Once each truck is near it's destination, it will peel off of the pack and park itself near the highway, at which point a real driver will hop in and finish the route.
I work in the telematics space, and work with extremely large fleets that you'd all recognize on the highway every day. None of them are afraid of self driving trucks disrupting them any time soon. However, these automated caravans could be their first opportunity to restructure the flow of their capacity across the country.
Too bad that all the "speculative VC dollars" seem to be in trying to jump directly to full-automation of multiple small passenger vehicles.
Now, I've never heard any public figure say this out loud -- so maybe I'm just barking up the wrong tree -- but I think the bandwagon is being pulled by a very particular idea about a market they can try to corner.
Specifically, they're thinking of a future demographic bump in people who are too age-impaired to safely and legally drive themselves. Folks who can't (or won't) pay another human to drive, or move to a location with better public-transit, but who still have enough affluence for a self-driving car.
IMO it explains the bubbly frenzied focus from multiple companies, despite the significant degree of basic research still required.
Part of the allure of self-driving cars is drastic reduction of traffic deaths. Long-haul drivers are not the one who kill themselves on the road the most; regular people are.
I'm not saying this is the driving force behind focus of companies, but it's an important consideration for many individuals in this space.
> The most likely thing (imo) that will happen will be highway trains
It already happens according to legislation. Though, these vehicle are generally limited to 2 trailers, restricted to some loads), if road condition allows it.
(Not being a jerk) if you just Google "truck telematics" the front page (and ads) are basically the big players. There are many more who you'll never hear about too.
They generally track the Tractor (truck) and Trailer, and also provide regulatory (hours of service) and Dispatch (sending jobs to drivers).
They also do a million other things, but that's the gist.
it seems they would need to build additional infrastructure. its my understanding that truckers would already like to caravan behind one another in very long chains (save lots of fuel), but its illegal due to traffic and safety concerns for other drivers.
"Then what you have is a truckers’ lounge with 20 or 30 guys standing around getting paid."
THAT's the error: those guys aren't getting paid anything to stand there. They get paid by the mile. Making them wait is benefiting the warehouse, and costing THEM.
Yeah. My wife has a friend who's a long-haul trucker. He gets profoundly disrespected by the current system, including showing up at 8AM to an appointment and being made wait all day, and sometimes into the next day, for his truck to be unloaded/loaded and ready for him to take.
And he's paid nothing for that. There's no penalty at all for wasting the trucker's time. I cite this as evidence that the truckers have no effective lobbying body, and therefore will in fact lose most of long-haul jobs over the next 15 years.
It looks like your wife's friend needs a better broker because all the companies we deal with charge us.
Our shipper won't schedule a truck unless we know that the load is ready due to how steep the fees can be.
When we shipped some equipment to a major trade show, they try to schedule it so that the truckers have a minimum wait but the show will also pay the detention fees back if they were the source of the delay.
Honestly, I would guess you've nailed it: The delivery company is collecting a fee because they are only paid for per delivery, so making a driver (and truck) wait doesn't otherwise make them money.
But there's no guarantee that the fees are passed on to the driver. Just in case, I'll pass a question back through my wife to see if some truckers are actually paid for waiting. They talk to each other; I bet they all know if anyone is paid to wait.
Depends, do truckers have a strong political lobby? If no, they'll be in trouble. If yes, then they'll be guaranteed to exist for as long as they are funding lobbyists and politicians.
In the United States at least, it's going to be devastating. I don't see a bunch of individual owner-operators banding together to protect their livelihoods. As soon as it's cheaper to run automated trucks, they'll get picked off one by one. Probably the best than can be hoped for is laws mandating a person as co-operator in case of emergency. There's also some version of the last mile problem, where people need to be present.
For fully automated trucking, I imagine no one has explored the security implications yet. These vehicles seem far too easy to hijack, redirect, or delay and steal the cargo. Take out the human factor and long-haul trucks are autonomous warehouses roaming the countryside.
I could see the last mile problem for trucking being solved by depots on the outskirts of cities where the automated trucks arrive, and human drivers commute and drive the trucks from the depot to their final destinations in the city. I don't know about the economics of it, but it could potentially be beneficial to the health of the human drivers as they wouldn't need to be making long-haul trips, but instead navigating more, but shorter, trips into cities. If this scenario increased the volume of long-haul trips (through automation), it could create demand for more drivers for the last mile trips.
Autonomous trucks may actually be a little more secure. The vast majority of truck thefts occur within 200 miles of the origin and truckers are strongly encouraged to not stop in this zone. Many still do though, either for legitimate reasons, or they're working with the thieves. Automated trucking will probably fix this problem. You can have automated trucks moving from secured site to secured site without worrying about exceeding a truckers 11 hour drive limit.
Aside from thieves, there's likely to be some mischief from the truckers whose jobs were taken away, or soon will be. They are already a group that's very tight, with lots of communication methods (CB, forums, newsletters, etc). And pretty good knowledge on how to disable/vandalize a semi truck.
I'd also expect more "insider" help than usual. Probably many truckers will be transitioned to lower paying roles and will be bitter.
I don't think exploiting software, in the usual sense of the term, is even necessary. Just box the truck in somehow when it's off the grid and can't phone home. There's no way the designers will build in the capability to ram through a barricade, will they?
My point, which I stated poorly, is that automated trucking is a whole new class of problems from a security standpoint. I'm leaving some scenarios out deliberately. I haven't even given much thought to the problem, but I hope someone does.
What do you think a human driver would do if they were faced with a barricade? They'll hand over the keys. Act quickly and they probably won't even have the chance to call the cops before you take their phone.
An automated truck will lock itself down, and start taking high def video. Plus, if it has a satellite hook-up, it can contact the police even from outside of cellphone range.
Laws would have to change with a person in the cab you are talking about Grand Theft, Kidnapping, and Robbery. With an AI it's just Grand Theft meaning you'd spend much less time in prison.
Why would laws have to change, after all there is no kidnapping and robbery occuring. Stealing the contents of an ai-truck is less terrible than robbing a truck driver at gunpoint. There is a reason one is punished more than the other.
It seems like it'd be far easier to rob a human-driven truck. Point a gun in the driver's face and they'll do whatever you want. For a machine, you'll actually have to hack into it.
They can also lock down the cargo in such a way that it sort of becomes a sort of totally-sealed (automated) roving container. Much like any security, there would be limits into how cheap they can make the container still relatively secure of course (those hijackers are pretty wily and clever!)...but that doesn't seem to be such a logistical challenge in my mind.
Insurers, in stark contrast to VC funded startups, are very conservative. I'm confident they'll rather not spend time and money now exploring something that's not going to be reality for at least half a decade, maybe ever.
I agree, they are among the most conservative. But I think that's why they are studying it. Insurance companies hate being surprised.
Self-driving vehicles will going to prevent a lot of collisions which will in turn drive premiums way, way down. The need to insure unattended shipments is one mitigating factor that I bet they are thinking about.
Given that truck driver is the most popular job in the majority of US States, I get the feeling its going to go beyond having a political lobby. You put that many people out of work in one event and that's going to wreck havoc. Political lobbies are not the only way to exert power.
Truck driving is not the largest employment by major NAICS job classification. Instead you'll find that the ranking puts office & administrative, sales, and food services ahead of all transport. The final column is per-mille, not per-cent, contribution.
43-0000 Office and Administrative Support Occupations 22026080 156.881
41-0000 Sales and Related Occupations 14536530 103.537
35-0000 Food Preparation and Serving Related Occupations 12981720 92.462
53-0000 Transportation and Material Moving Occupations 9731790 69.315
51-0000 Production Occupations 9105650 64.855
25-0000 Education, Training, and Library Occupations 8636430 61.513
29-0000 Healthcare Practitioners and Technical Occupations 8318500 59.249
47-0000 Construction and Extraction Occupations 5585420 39.782
49-0000 Installation, Maintenance, and Repair Occupations 5456640 38.865
39-0000 Personal Care and Service Occupations 4514960 32.158
37-0000 Building and Grounds Cleaning and Maintenance Occupations 4426090 31.525
15-0000 Computer and Mathematical Occupations 4165140 29.666
31-0000 Healthcare Support Occupations 4043480 28.800
33-0000 Protective Service Occupations 3386360 24.119
17-0000 Architecture and Engineering Occupations 2499050 17.799
21-0000 Community and Social Service Occupations 2019250 14.382
27-0000 Arts, Design, Entertainment, Sports, and Media Occupations 1902970 13.554
19-0000 Life, Physical, and Social Science Occupations 1152840 8.211
23-0000 Legal Occupations 1075520 7.660
45-0000 Farming, Fishing, and Forestry Occupations 463640 3.302
I didn't downvote you. I'm guessing it's about two things.
The parent said the majority of the US states. [1] The premise there is - based on the data I've seen at least - in many states trucking is the most common specialized job. Sales and office jobs are non-specific as a super category. It doesn't make a lot of sense to compare something as extremely vague and varied as "sales" to trucking. Food prep & serving, does have some claim on the most common job, although that's generally a much lower tier of work in terms of pay and upside potential (fry cook, cashier or runner at McDonald's vs making $80k a year long-haul, not very comparable skill-sets).
And you didn't go any deeper on the numbers in response to large differences between states in employment (given the parent proclaimed "in the majority of US states"). For example, in some states is transportation and material moving a more common job than food prep, and which states are those? (etc etc)
[1] According to NPR's Planet Money, truck driver is in fact the most common job in most states. They reach that conclusion by dumping the overly vague super categories such as sales.
The detail level I provided is the "major" classification division. At "minor", trucking is only the 12th largest employment classification. There's a breakdown to fine detail (and I'm not sure what classification system is used) in which "truck driver" is the most numerous single entry, but that's almost entirely an ontological artefact: truck driving doesn't have a bunch of fine subdivisions. Contrast the employment classifications for "teacher", which has something like 50 distinct sub-categories.
1 41-2000 Retail Sales Workers 8791750 62.619
2 35-3000 Food and Beverage Serving Workers 7355090 52.387
3 43-4000 Information and Record Clerks 5630810 40.105
4 29-1000 Health Diagnosing and Treating Practitioners 5143640 36.636
5 13-1000 Business Operations Specialists 4629810 32.976
6 53-7000 Material Moving Workers 4580950 32.628
7 47-2000 Construction Trades Workers 4216890 30.035
8 25-2000 Preschool, Primary, Secondary, and Special Education School 4133490 29.441
9 43-5000 Material Recording, Scheduling, Dispatching, and Distributi 4101520 29.213
10 15-1100 Computer Occupations 3997370 28.471
11 43-9000 Other Office and Administrative Support Workers 3940510 28.066
12 53-3000 Motor Vehicle Operators 3934070 28.020
13 35-2000 Cooks and Food Preparation Workers 3190940 22.728
14 37-2000 Building Cleaning and Pest Control Workers 3174220 22.608
15 43-3000 Financial Clerks 3133030 22.315
16 29-2000 Health Technologists and Technicians 3018820 21.502
17 49-9000 Other Installation, Maintenance, and Repair Occupations 2836540 20.203
18 39-9000 Other Personal Care and Service Workers 2820750 20.091
19 13-2000 Financial Specialists 2651370 18.884
20 51-9000 Other Production Occupations 2582350 18.393
I've looked at this before ... in June of 2015 ... and the result was very nearly the same. From then:
And yes, I have not broken this down by state, let me see if I can find that data. It's also worth noting that the NPR article uses Census classifications, which may differ from those I've listed (though Census is one of the participants in the NAICS classification, itself shared between Mexico, the US, and Canada).
NPR are ... unclear in the extreme ... as what specific data fields and employment classifications they are using. Given the NAICS categories above, multiple states would have to show peak employment in one of the listed categories above "Motor Vehicle Operators". And yet, none do.
Truckers are mainly small businesses so they probably have no lobby. The bigger trucking companies that employ these self-employed truckers will happily switch to automatic trucks. They also have the political clout.
Same for farmers. The real small family farmers have no political power. It's the big farms that get laws made for them.
As a small farmer, the government requires that we pay into an industry lobby group each year. All farmers, that is, not just small ones.
But, as a small farmer, I already see how unaffordable automation is. Tractors that are basically self-driving have been commercially available for decades, but it is difficult to pencil out unless you have a lot of acres to spread the costs across. I feel that the same will be true for self-driving trucks. The huge fleets will adopt it immediately, but small guys like me who only have to haul grain a few weeks of the year, it seems more cost effective to just spend some time in the seat.
Time will tell, but history tends to be a good indicator. I don't see truckers going away any time soon.
Everyone talks about truck drivers becoming obsolete once self driving cars take over the road.
What I would worry about are endless number of doctors and surgeons that spent a life time mastering a skill and will just as easily be rendered obsolete by the same technology.
Don't worry about doctors. Political power wins out in the end, and they have tons of it.
For example: for decades they've managed to resist entry of decision-support systems(which were not perfect, but still, an improvement), even though their own error rate is ~1/8.
Well actually its economic power. These people just have to pray technology doesn't get cheaper with time.
I remember in the 90s an entire generation of Indian workforce lost jobs overnight, for decades they fought computers and software. They bought in regulations, laws and what not. Unions fought 'computers' tooth and nail. Yes, you read it right- they were against the very use of computers.
When the prime minister called for use of computers in India in the 1980s there was a national strike.
Eventually it became impossible to not use computers without slipping into an economic stone age. People had to give way.
A couple of years back, after being tossed around by a dozen doctors for simple infection. A few hours of internet research helped me solve my problem, better than any doctor did.
Health care industry is a cartel that optimizes for profit.
> What I would worry about are endless number of doctors and surgeons that spent a life time mastering a skill and will just as easily be rendered obsolete by the same technology.
You mean, because they operate on car crash victims?
The Jevons Paradox implications of reducing the costs of litigation by one or more orders of magnitude ... terrify me. The more so as reducing costs tends to increase the level of concentration of power.
Mu impression is most commercial flights are largely run on autopilot. AFAIK, pilots typically only get involved for takeoff, landing, flight path changes and abnormal events. If last-mile piloting was a thing, I'm sure the airlines would try to cut costs by getting rid of the crew for the auto-pilot segments. It seems trivial for an automated truck to pull off the highway and wait for a trucker to deliver a vehicle, yet impossible for an aircraft to hold at 30k feet and wait for a pilot.
A huge difference is that when something goes wrong with a truck then the autopilot can just apply the brakes and let the truck come to a halt and wait for help.
>It seems trivial for an automated truck to pull off the highway and wait for a trucker to deliver a vehicle, yet impossible for an aircraft to hold at 30k feet and wait for a pilot.
I think many people will not like the idea of sharing a highway with an unmanned truck.
I've done my fair share of driving (including a couple cross country drives to SF for internships), and the scariest parts usually involved passing or being in close proximity to semis.
It would be sort of cool if highways could then promote a "goods carrier trucks can only run nightly" rule with this. Of course allowing essential goods trucks to run in daylight
> while truckers are prohibited from driving more than 11 hours per day without taking an eight-hour break, a driverless truck can drive for the entire day. This effectively doubles the output of the trucking network at a quarter of the cost.
Interestingly, doubling the throughput of the trucking network would also halve the average lifetime of road surfaces. "The federal government has estimated that a 40-ton, 18-wheel truck causes the same damage as 9,600 midsize cars."http://www.latimes.com/opinion/readersreact/la-le-0215-sunda...
The automation of vehicles is taken to strongly imply robotic devices elsewhere in society.
Apart from more sophisticated robot hoovers, lawn mowers I don't think that is necessarily so.
Hoovers and mowers are safe because they are small, light, low to the ground and constrained to floors and lawns.
Vehicles, fast, heavy, high, are in a controlled environment.
Other controlled enviroments like factories have already ventured into robotics.
It seems to me that main advantages of automated movement would exist in unstructured environments like streets, shops. The robots I've seen so far have low center of gravity, are heavy so as to inhibit any danger.
What exactly does automated driving do for other unstructured environments?
It is far from clear to me that this is going to be a trifle soon after the roads swarm with autos.
We don't even have autonomous street sweepers (well defined environment, not allowed to run over stuff) mining dump trucks (not as well defined environment, few things it's not allowed to run over and most of them have reflectors on them) or forklifts to unload trucks (well defined environment, nothing to run into).
I haven't thought of this until now, but it's going to be interesting to see the impact of self driving trucks on human migration. Today many people feel trapped where they live because it's prohibitively expensive to move. Once self-driving trucks are common, the cost of moving across the country should plummet. Imagine being able to move everything in your apartment or home across the country for a few hundred dollars.
The only move I've coordinated cost quite a lot. It involved moving a 4 bedroom house from the South to New England and cost about $4-5k with labor and boxing and only unloading at the destination. The majority of the cost was the labor after everything was boxed and loaded.
If you're loading and moving yourself, a U-Haul or similar is not that expensive. I've done interstate moves and the truck rental was never more than a few hundred dollars.
True, but there is another cost there, opportunity cost. The week you take to drive cross country is a week that you aren't working and earning money. You're no longer paying the truck driver for his time, but you are now using your time. Overall, this self-move approach works for short moves of maybe 1-2 days time, but the moment a move will take more than a weekend, you're disrupting your earning opportunities.
With self-driving trucks, you will be able to move for a few hundred dollars and be able to focus on employment.
The freeway is where AI will be first. It is the most controlled driving environment we have. It can be made even more controlled in a fairly efficiently and cheaply. The local city and country roads have a plethora of other issues. Pedestrians, driveways, trees, multitudes of entrances and exits...its a nightmare really.
Give me AI on the freeway, and I'd probably be happy enough.
Automation is one thing in fully mapped Mountain View with perfect weather. It's a whole other job when supply lines go through rough terrain, like Canadian western highways in winter... those chains won't get around those tires by themselves, not to mention the risk of mechanical failures.
Just the threat of automated long haul trucking should cause rail freight rates to decrease, ultimately rates will drop dramatically. Last mile is the final holdout for truckers.
If the Starsky model ends up dominating, this could all add up to a case where the price of truck transportation drops so much that demand for truck drivers actually increases, possibly even counterbalancing the pressure of individual truckers being able to manage multiple vehicles at once. This would be an example of the "Jevons Effect" https://en.wikipedia.org/wiki/Jevons_paradox