Truckers Working Alongside Coders Trying to Make Driverless Trucks
bloomberg.com
bloomberg.com
This is amusing because this stopped being true a long time ago. Even by 2007, it was hard for anyone to improve, and after 2013 or so, the very best centaurs were reduced to basically just opening book preparation (itself an extremely difficult skill involving compiling millions of games and carefully tuning against the weakness of possible opponent engines), to the point where official matches have mostly stopped (making it hard to identify the exact point at which centaur ceased to be a thing at all).
If you think about it, AIs have a significant advantage in that their evolution is completely artificially guided. They can be focused on tasks in isolation, and the results can be observed and improved upon. Biological evolution does not have this luxury, overall the entire organism has to "work", and survive long enough to reproduce. Sure, current technology is in its infancy, but extrapolating into the far future, things are interesting.
It's also interesting in that it demonstrates that there's at least two levels of 'superhuman' performance in chess and other domains: there's superhuman in being able to beat any living human one on one (Deep Blue, Sedol AlphaGo), and superhuman in being unable to be improved upon by a human collaborator (post-2013 chess engines; AlphaGo Master?). I'm not sure this gap was really expected, but though it only lasts a few years (or months), it's still interesting to note.
You get paid to train your replacements. If you don't train them, someone else will. It's a no win for them. They don't have pull or unions that commuter train operators have so it's not like they'll get to ride along and supervise the autonomous vehicle, in the long term. Instead, there may be an ops center somewhere in TX where operators intervene remotely when an irregular operating pattern pops up. Those will be fewer operators.
Wages are a trucking firm's biggest expense. Why would a firm go out and buy a bunch of expensive self-driving trucks, so that they could pay the same number of people the same wages.
Halving the cost of transporting goods will not double the demand for transporting goods. Consumer budgets are fixed, and transportation is a very small fraction of the price of goods that consumers buy.
Dropping the cost of transport by 30%, because you only need to employ half the truckers, will not double demand. Unlike the before-after world of trucking, air travel, container shipping, etc, there isn't a mountain of businesses that suddenly become viable when long-haul costs get slashed by a third.
What fraction of the economy do they make up?
What if efficiency gains are not 2x, but 10x? What if you'll only need to employ one trucker to do the job of 10? Do you think demand for trucking will go up 10x? This isn't a web framework.
Trucking costs are built into most products. My grocery store sells apples from California. If the cost of trucking them to me goes down, those apples will decrease in cost and they will potentially sell more apples.
Automation of driving is all about making people with lots of capital richer. Efficiency is just the dream -- this stuff will dramatically increase costs of shipping as capacity is constrained.
For example, I just requested quotes for waterjetting some stuff today. If LTL shipping were cheap enough to I'd look outside the high CoL sphincter of the country where I happen to reside but instead I'm stuck paying a premium because the local cost of everything is high. There's also a cylinder head in my garage for a 60s Ford I'd like to give to someone who needs it but nobody needs it bad enough to pay for me to ship it across the country.
At the higher volume levels similar stuff happens. A construction company may source steel from farther away to undercut it's competition. You can bet the logistics people at Walmart (and anyone else with a big fleet) are drooling over all the optimizations they could implement if trucks could run like call centers.
This is the only point I seek to make.
If salaries are 60% of the cost of shipping, and self-driving trucks make drivers twice as productive, costs will drop 30%.
This will bring a demand increase of less then 30%. Let's say 15%.
... Doing the math, it looks like 42% of truckers just lost their jobs. But at least they can buy more apples with the money they no longer have.
The bigger the productivity gain, the worse it will look for them.
Lots of products have price elasticity of greater than 1.
Shipping seems like a very elastic industry. If anything, demand would increase by more than 1% for every 1% decrease in cost.
This is because shipping costs if shipping costs are low enough, then you can build something for cheap somewhere else and then ship stuff across the country to high cost of living locations.
YOU wouldn't ship stuff more, but industry would.
IE imagine if something costs 1/3 if it is made in China.
If shipping costs are more then you save, then nobody would ship it. But if shipping costs go down, now you can easy get those cost benefits of across world or across country manufacturing.
Possibly since not all my cars are electric yet.
You can also see this happening for other resources
On the other hand, if the price of that ticket dropped from $400 to $360, we wouldn't see a 10% increase in passenger volume. The set of people who will fly somewhere for $360, but won't for $400 is tiny.
I posit that trucking is in the latter, rather then the former category. It's already cheap.
I disagree. Given companies like Amazon, it is currently not cheap enough. I would argue that it will take longer before it's 'cheap'. The appetite for 'free shipping' is still growing and it's very different from the demand for airplane tickets. Kind of like with your electricity example, you're missing parts of the big picture.
But what percentage do trucking expenses make, of their overall sales receipts?
Will their sales double if the cost of trucking drops by a third?
Yes I could be wrong but I feel further lowering the costs of long distance trucking would help increase eccomerce sales for retailers which in turn would further increase demand for it
Cost stay the same, but trucking capacity triples.
Yes wages are the biggest expense but I think the big pushed for self-driving trucks is 1) Reduced liability 2) Trucks can be operating nearly 24hrs a day. 3) Greater fuel efficiency.
So sure some will lose their jobs but I think half is overly pessimistic.
In comparison to a computer programmer, yes. In comparison to the nationwide average, no. https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&c...
I can immediately think of one (retired) truck driver I know who comfortably raised a family of 6 on a single salary.
https://www.usatoday.com/pages/interactives/news/rigged-forc...
Most of those roles managing bullshit at the head end will go away as the robot fleets consolidate and the current carriers die off. Customers will have to take or leave on whatever terms the carrier demands -- just like the old railroad days.
There is little basis to declare human drivers 'fundamentally unsafe' given the proportion of accidents pales in comparison to the gigantic volume of vehicles on roads worldwide over varied road and traffic conditions.
It can be speculated automated vehicles may require constraints on roads in which case it becomes apples to oranges. And human drivers equipped with the kind of technology and sensors on self driven cars would be much safer.
While computers lack human weaknesses, they bring a slew of new ones. Computers have no situational awareness, no context, are vulnerable to hacking, respond in predetermined ways which can be manipulated or abused, etc.
There just is not enough evidence to make the claim that ubiquitous autonomous cars will be any safer than human drivers on shared roads. There are many known unknowns and who knows how many unknown unknowns left to discover and mitigate with this tech.
https://techcrunch.com/2017/03/17/laying-a-trap-for-self-dri...
Today that is true. I am speculating when I say that autonomous vehicles will be safer than humans drivers. There is enough effort going into developing them that we will soon have that data though.
Google's data is already showing promise that the autonomous car is likely to be safer, but they don't have enough miles to be significant.
> There is little basis to declare human drivers 'fundamentally unsafe' given the proportion of accidents pales in comparison to the gigantic volume of vehicles on roads worldwide over varied road and traffic conditions.
10.6 deaths per 100,000 people is enough for me to call humans fundamentally unsafe. (though cancer and other medical conditions are bigger)
Since humans cannot be equipped with the sensors a computer can (ie eyes on the back of your head) it isn't fair to compare the as if situation.
You don't think so? Seems like an unmanned truck would be a great target for theft.
That's true, but the problem is which one? Most of them are dry vans that you don't know what's inside, and/or not worth the risk. My loaded trailer was once opened and they didn't touch the merchandise. The only loss was the lock.
Just go on an open stretch of road in the country side with a firearm and open fire through the truck windows. A truck driver isn't going to be able to defend themselves.
You might argue that this doesn't happen because criminals are afraid of getting caught by the police and sent to jail.
But, if anything, the risks would be even greater for a self driving truck that has dozens of always on cameras recording, GPS tracking, and an emergency police call system.
So by the time a truck is stopped, realizes a robbery is in progress, contacts command, who assesses if it's a real event, who contact local police, who send a cruiser to check investigate. They thiefs could have gotten away with quite a lot.
Btw, a grand theft rarely rises to immediate action. City PD have bigger concerns, most of the Midwest is sparsely populated. That leaves the bored suburb PD who'll be in the most responsive.
2 aspects
1) If drivers are at such danger that they do more than contact police then the liability would require they have trained security with them.
2) Drivers themselves are a major security vulnerability, how susceptible is a driver to being paid off by a "pirate". [and there was a recent article in Outside about how healthy nuts go through a lot of truck theft]
I would compare it to banks. ATMs are still a target of theft but it's much harder to crack them compared to robbing a bank where tellers have access to cash. The same will be true for trucks. There will be some thefts that are more sophisticated (and successful) but the overall rate will go down.
Semi-autonomous driving of trucks could change the job for the better. It would cost a lot of jobs but then again, few people become truck drivers because they love being on the road for days. For most it's a job and they'd be happy to take another one. And there is currently little evidence that no new jobs would come up as they always have for past disruptions.
Driverless trucks are a very interesting concept. Once they're on the freeway, I would think most automation would work (relatively) well. But getting through traffic congestions, busy multi-turn-lane intersections, and making sure you don't sideswipe someone who's hugging the line seems rather complex given the different trailer dimensions and blind spots, (un)marked lanes and everything else it takes to get to the highway.
I would think that a physics model would also give a reasonable automated system superior control over where the trailer is at a given moment (and to account for different trailers).
That's assuming nice weather conditions. Mudslides, flash floods, extreme code, ice, snow, etc. needs to be taken into consideration.
And if there _are_ dangerous conditions that a human driver could navigate but an autonomous one would give up on, is that so bad? The industry is making enough efficiency gains to offset that.
I'd imagine the quality of life would be significantly better too since they would work in the same spot with co-workers and could go home every day.
Actually, working in the same spot with coworkers was exactly the reason I started driving a truck. Going back to that would be an enormous downgrade in quality of life for me and a lot of drivers.
The people duped into this will lose their driving skills and pretend it's other humans fault for not conforming to the program they invested in. I'm not all that worried, the market will sort this out in the US at least.
Can't even imagine a manslaughter case where a driverless vehicle killed someone. Who's at fault? Most fault current resides in the driver, as they're the final call.
Exactly. Humans. Humans are the bugs in their code. The control writers will argue for more laws so their programs work. The possibility that their code revisions are futile will dismissed as [insert buzzword]. I am excited that we get to watch this play out. The wildcard is biological CPU's: http://www.research.ufl.edu/publications/explore/v10n1/extra...
Trucks getting cut off is the norm in many areas. Punishing people for normal is generally not smart.
Drug smugglers are another group that could use this and how would a cop stop an autonomous vehicle to search it they would have to have a back door which could be abused
Outside of national borders, smuggling felonious quantities of hard drugs from state to state is already trivial.
> Ok, but what if the refugees were lowered by a helicop...
Sensor on the roof. Sensor on the side. Sensor everywhere.
My brother is a lorry driver who frequently crosses the Turkish - Bulgarian border with his truck (as in once every 2 weeks). Once or twice he caught refugees (or whatever you want to call them) trying to get inside his trailer by cutting through the thick textile material that covers it, while he was awaiting at the border. You generally do not whant people mixing with your industrial equipment while the truck drives at 90-100 km/h. In other cases it happened to him that other people had stolen diesel gas from his tank while he was asleep during the mandatory break (this happened somewhere in Western Europe, I forget where).
And leaving aside all this, assuming that bad people won't suddenly stop stealing and generally doing bad stuff to unattended expensive material (like trucks and the things that the trucks carry), being a truck driver you're also doing lots, and lots of bureaucracy work. I once shared the cabin with him as a passenger during one of his routes, while we passed two European borders. The amount of paperwork he had to fill in was mind-blogging, I half-jokingly told him that his job looks more like a clerk-job than being an actual driver. Now, lots of clerk jobs have already been automated, I know that, but it mostly happened in the private sector, inside private companies doing business with other private companies. But once your clerk-like job involves having to do business with several State-like entities things become a lot more complicated, with legislation changing once every 6 months, being very diverse from State to State etc.
What I'm trying to say is that theoretically the self-driving problem looks achievable for trucks, but once you get into the messy details things become a lot more complicated.
He should let them ride with him in the cab up front, then.
It's the human thing to do - and in another universe (or if this universe were configured only slightly differently) they could be his own family, after all.
They are not going to let you cross the border with a cab filled with refugees/asylum seekers/undocumented immigrants.
The point is, I'm actually amazed at the downvotes, and that people aren't naturally, viscerally having a more human response -- "For heaven's sake, can't I think of a way to help these people? They're fearing for their lives, after all!" -- rather than the response described in the post I was responding to: "Oh shit, it's risky/complicated! Avoid, avoid!"
Especially given this history, among other sordid events in the past:
And to think that just because Turkey isn't currently running gas chambers that these people have nothing to worry about is, unfortunately, very naive:
Amnesty International has slammed Turkey for expelling dozens of Syrian refugees to their war-stricken country, saying the move is in violation of international law.
According to a new report released by the UK-based rights group, a total of about 80 Syrian asylum seekers, who were previously held incommunicado in a detention center in the Turkish city of Erzurum, were forced to leave for their country, with 50 more people facing deportation.
-- source: look it up yourself
I also have to imagine they will be a ripe target for vandalism, especially if operators don't aggressively remove graffiti.
And they'll also be surrounded by camera's and hard drives. I would imagine that the sensors that can see through fog could also get more detail than the average CCV camera - meaning that even with a mask, you're going to get a lot of detail on vandals or road hazards (aka: You going 5mph in front of one)
Is a lone driver with a tire iron, mag-lite, or billy club really the deterrent to this? Especially, the stereotypical obese early 50s driver?
None of that will change and that's not dependent on drivers.
Car drivers already take advantage of trucks leaving a safe distance or their slow acceleration in traffic.
Humans get sleepy, get emotional, zone out, and make all kinds of mistakes all the time.
The first generation of driverless trucks will still have a human as passenger to be backup in case of mistakes. You may be thinking, "What's the point if there will still be a human there? s/he might as well drive." Well, humans still need breaks and can help give feedback to Truck for it to improve.
The first generation won't be perfect, but it will be on its way.
AI sensors combined with some sort of 360 view VR with drivers for the last mile with full AI on highways/freeways would allow for progressive roll out of self driving technology. A lot of video to push down the wire though but probably manageable with modern codecs.
Perhaps this is more of an occurrence on re-treads.
It's good to see someone doing more than just talk about this sort of thing; both the employment aspect and the culture clash side of it. Compare that to say, how Google treats its ex-military drone contractors [1].
1. http://www.businessinsider.com/ex-military-contractor-google...
> "We hire truckers," Seltz-Axmacher tells prospective drivers right before offering them a job. "But we also have a lot of engineers in Silicon Valley. Everything you’ve heard about San Francisco—it’s all basically true. There is something called raw denim, and in San Francisco people wear it, which means that some of your colleagues will pay up to $300 for a pair of blue jeans. They sometimes drink $7 lattes, too. Many of your co-workers will not be from the U.S. They will have accents. Can you handle that?"
In my experience, the culture clash seems to me to be more common in technology companies doing hardware projects than in hardware companies adding software to their processes. As a member of the latter, I take exception to this statement:
> “We basically have people from two worlds, neither of which has ever talked to each other,” says Seltz-Axmacher, who grew up in suburban Maryland. “That’s kind of what’s wrong with this country.”
Perhaps that's true of the founder's life experience, or true of the trucking industry, but most manufacturing jobs seem to be merging these cultures comparatively smoothly. I've been at a couple companies in the automotive industry, and been to hundreds of different plants, and everywhere you look there are blue collar technicians running the machines and white collar engineers building and automating the machines. The situation is not so bad as the article paints it outside the extremes of Ivy League PHD programs contrasted with the trucking industry.
> ...culture clash seems to me to be more common in technology companies doing hardware projects than in hardware companies adding software to their processes.
The culture clash is Silicon Valley banging its own cymbals.
That's what I was thinking. They might be exaggerating a little bit the difference in culture.
I'm surprised they don't send their engineers through a truck driving course, and have them get a CDL. If you're going to work on self-driving truck software, you need to be able to drive a truck. At least get a Commercial Learners Permit.
That's nice to see, since the change-the-world spiel is almost always howling bullshit.
For those that haven't read it, the book is an interesting and prescient story about what the future of America looks like when nearly all human labor has been made obsolete.
Me too. There's so much more than just driving. Just as an example, we truck drivers learn to anticipate possible problems based on other people's behaviors. Why anticipate? Because 80,000 lbs can take too long to stop, and people can die.
I hope this is a joke. If someone literally cannot handle working alongside someone who voted for Trump, that is so sad.
I still think driving a truck as a profession will be a thing of the past within 20 years or so. There's just too much money riding on this problem, and logistics companies are fairly ruthless about efficiency.
I hope so. Trucking is one of the toughest professions for a person to endure that I've ever seen. It destroys a person's body and mind...and you're lucky to end up making minimum wage when you account for how many hours they spend working.
And yes, the government. The funny thing about a democracy is that if enough people want the govt. to provide something, it will.
If you're driving fleet, that's correct; however, there are a large number of "Owner/Operators" out there and they do make significantly more than you'd expect. If you have a good freight contract and a logistical mind, an O+O can easily make around $125k/year.
And there will be major accidents with multiple deaths, since the average passenger car driver is so unskilled. No software can account for every idiotic move made by the typical car driver, and big trucks take a lot of room and time to stop. A few lawsuits later, all the financial gains will disappear for the trucking companies and drivers are back in the seat.
I could see in 20 years a system where computer assisted driving automates much of the trip, much like commercial aircraft operate now.
Yeah, city driving is hard. That might be a long ways away. But highway driving? Thats easy. You can even buy a highway driving, consumer self driving vehicle right NOW. Tesla sells them.
And while highway driving generally has fewer decision points per mile than city driving, the speeds are much higher and so the effects of errors are much greater on the highway. The worst wrecks I've seen have all been on the highway, not in cities.