Popular Mechanics reports Self-Driving Freight Truck Just Drove Across the Country to Deliver Butter.
Plus.ai, an artificial intelligence startup in Cupertino, California, has engineered an autonomous driving system for commercial freight trucks. This week, it made the world’s first cross-country trip of its kind to deliver butter to a small town in Pennsylvania.
Key Details
- The Plus Ai truck drove from 2,800 miles from Tulare, California to Quakertown, Pennsylvania, through tunnels, snow, toll booths, and mountains, without a backup safety driver ever taking control.
- The truck relied on Plus.ai’s SLAM technology, which is an acronym for its instant positioning and map building solution. SLAM includes a suite of sensors like cameras, lidar and radar. What stands out is the company’s data fusion system which combines this information to create a field of front detection that’s over 1,600 meters deep, allowing the truck to see far ahead. Plus.ai also achieves a wide field of view to help the truck adapt to new road shapes and slopes.
- The truck was hauling a refrigerated cooler with 40,000 pounds of Land-O-Lakes butter onboard. This was a real delivery.
- This trip took place during the heavy traffic week of Thanksgiving, encompassed 12 states and some inclement conditions, including snow.
Adoption Timeframe
Despite this being 2019, I will still have readers telling me commercial operations won’t be in place for a decade, that drivers are better, and for numerous reasons these deliveries can never work.
Once again, I expect to hear the same tired arguments about truck hijackings, mountains, balloons, weather, cities, etc.
My position is that within a couple years of Federal approval, mass adoption of autonomous will ramp up quickly, for point-to-point interstate driving.
Note that this delivery went to an actual final destination. But a driver will take over to make an inner-city delivery, if necessary.
Cost and Time
Haulers with drivers will not be able to compete on costs or time.
Read that line over and over until the message sinks in. Competition ensures mass adoption on an accelerated timeframe.
All we are waiting for is federal regulation.
Early adopters will come on within a year, then the rest of the industry will have to adopt or they will be soon be priced out by market forces.
Mike “Mish” Shedlock
According to this Mckinsey report, Autonomous Trucking could roll out in 4 waves
You can self-drive a truck thousands of miles across the country, but delivery to your door will be done by your friendly Indian delivery guy.
I hope this succeeds and this is a step forward, but Mish is a long way off from claiming victory. For all we know, the software may have been written specifically for this one route. They had a good demo. But sometimes going from demo to actual product takes years.
The next time an 18 wheeler is tailgating you at 75, using the rumble strips to nudge himself back in the lane, reflect on whether you’d prefer this AI system to the human behind you casually taking your life in his hands.
Insurance companies will take the drivers out as soon as the data show 95% reduction in preventable deaths and carnage.
I’d remind everyone that Henry Ford’s initial approach to using his little gasoline engine for transportation was a bust. He tried to sell it as a power booster kit on carriages in San Francisco. Ridiculous! Too noisy, scared hell out of some horses, and even if it worked, it would devastate the horse and mule business so important to the economy…
He accepted defeat, went back to Detroit and decided to try it with his own little carriage. It indeed devastated every mule barn and buggy whip firm in the country.
It’s coming.
People accepted change in those days. Now, there are always reasons why lawyers and vested interests will block progress.
large numbers of self driving trucks are still 5-10 years out: The basic problems haven’t changed.
Large chunks of the savings are illusory if the trucks end up legally required to have a backup driver on board. That driver will end up subject to labor laws and safety regulations restricting the time that the trucks can run continuously and will be paid reducing the savings dramatically. Given the extra costs up front costs of a dual system a single poorly (or intentionally) written federal law could completely negate the advantages of self driving systems.
Software is great for scaling, but also terrible for scaling when issues are discovered. A single problem in a single aircraft can threaten to ground every aircraft of that type, and the solution is in part a slower, steady approach to replacing aircraft, not wholesale changes across the entire industry. Again adoption will be slower because issues are magnified by volume here, and slow and steady will likely lead to the dominant players controlling the industry with fast changes being far more likely to collapse companies.
However Tort issues are basically non existent if self driving trucks are significantly safer than human drivers. A single huge settlement won’t matter because that is exactly the case that insurance is set up to handle, thousands of people are killed annually, and tens of thousands of people more are injured, in accidents involving trucks, you can easily handle a 10x payment per death/injury increase if you get a similar collapse in actual incidents.
Finally if you think that all we get out of self driving trucks is sightly lower shipping costs you don’t know the extent and costs of the trucking industry. Potentially 5,000+ lives saved a year and tens of thousands of fewer injuries from safer trucks, much less road congestion and less pollution with a less impact from what pollution there is. Perhaps we should have shirts saying ‘Self driving trucks stole our jobs and all I got was an extra decade with my granddad and this lousy T-shirt’ printed up.
For those who think 1 self driving truck will lead to 1 million self driving trucks over the next couple of years, here is what the founder of the self driving truck company says:
“The transition is actually slower than you thought,” he explained. “The transition is not like, ‘tomorrow, all of sudden, trucks will be autonomously driven.’
Mish, I must admit that — while I’ve read your previous posts about this issue — I’m a skeptic. However, I do have an open mind about the possibility of the wide adoption of self-driving vehicles.
IMHO, tort liability will have a big role to play in this field. If a self-driving truck slams into a minivan — killing or maiming a family — and a jury awards YUGE damages against the manufacturer & software firms — then the adoption will take decades. If the benefits of self-driving trucks become readily apparent & publicized, the adoption will happen mich more swiftly.
Surely a railway line for long distance haulage is much cheaper than even autonomous trucks?
You’d think so, but I’ll bet it’s about the same once you factor in all the transfer fees.
It’s not always about bottom-dollar cost, there is a time/cost equation where the length of time in transit is a factor also.
Highways are subsidized, which reduces the cost.
Out of curiosity what was the backup in case the truck met with an accident?
Whoa! 4 hours and the comments are sparse and trend to humor. And the “con” side is half-heartedly tossing in, “Gosh, that’s a lot of change – we can’t have that.” And “By golly, it’ll be expensive.”
Mish, Mish, my friend, don’t be so discouraging. We come here to learn how bags blowing across the Interstate are an insurmountable problem.
Mish is right though, you can’t argue with a machine because it has nothing to lose. That silence will get written in as consent by regulators.
SLAM? Someone made a funny.
The hot new consumer drone is the Skydio 2. It will follow subjects while avoiding objects with minimal intervention. And it does a pretty good job of making interesting pictures too:
Crosshairs are an extra though.
That’s an awesome idea, a drone buzzing behind you while you run or bike in nature. Should be multiplied by the number of runners for a full experience. Certainly worth a thousand bucks.
Mish,
I am completely with you on the interstate movement of goods with self driving vehicles. Computers are very good at repetitive task with little change in the bounties. Interstate fall into this category. They change very little from day to day and a route can be mapped out from one point to the next before the truck ever leaves. Lane changes road closures and even traffic and accidents would be factored in using google or another map service.
I would expect for this to be economic but a build out of transfer stations would be necessary. 30-40 of these across the nation around major areas would make transferring to human drivers the last mile very easy.
95% long haul is interstate so at least 75% of long haul trucking jobs would be lost. The rest would be last mile service to distribution center.
Once this technology gets cheaper than drivers the adoption rate will be breakneck.
A crash would have been very messy.
A slippery slope to disaster? 🙂
Driving butter 2,800 miles from the Central Valley to Pennsylvania? I do not want to hear one single word about man made global warming when there are THOUSANDS of dairy farms within about 1,000 miles of that destination.
As to COMPUTER driven vehicles, they are monstrosities and you are going to regret the day you ever heard of them. They are going to be the single greatest tool of your surveillance the government has in about 12 or 15 years. And please stop calling them “self driving,” they are not, they are computer driven and cannot function without the cloud, that is why you are about to be taxed between 4 and 5 TRILLION with a “T” to pay for 5G because without 5G there simply is not the scale of bandwidth to have widespread computer driven cars.
Are you sure? Your mobile phone already knows your every movement and can hear and record everything you say.
We’re learning, people can secure their phones but in a decade most people will still be reliant on a vehicle, and it will need to be switched on, connected, and not tamperable by owner to be allowed to be used in the case where it is computer assisted. However it would only really track certain movements, given that speech would be known to be recorded, so more limited but a barrier to privacy all the same.
How many people actually secure their phones? Almost nobody. And the phone follows you everywhere, even to places where a car can’t. Your home, your office, all business meetings, overseas vacations…
I do turn off location, but I seriously doubt that does anything. I’m sure my location is still tracked.
I installed Linux on my phone. Better than relying on Android/iOS settings. But the only real way to not get spied on is to not have a mobile phone. Then only everybody around you has spying devices.
Which Linux distro? Are iOS or Google apps running on it?
My current phone only supports Ubuntu Touch. I can run any Linux app on it. Some Android apps are possible but complicated.
I have ordered a new phone that can run Sailfish OS, which has a good compatibility layer for Android apps.
There are also Google free clones of Android, like /e/
I have GPS shut off on my phone, but of course they can still see what cell you are in even if they can’t determine where in that cell you are. I don’t have Alexa or any of the other home spies and will not be getting them.
Once we have no choice but to use ride sharing computer driven vehicles to run our lives I will be away to someplace that cannot afford the infrastructure those require. A place to go live out the remainder of my life where coffee is grown and nobody demands I stop tobacco. Where you do not need batteries to read a book. And you can still live without a computer or internet.
You can do though. For calls an unregistered dumb phone and foreign sim (unregistered) , I change mine every couple of weeks just because I get more internet that way than on plan. Have about 50 sim cards lying around. You can’t change a dumb phone identifier, but “they” don’t know it is “you” unless they physically trace the call location and grab your ID somehow. Anyway dumb phones are cheap to burn if you feel that is the case. Online use a good VPN, I use safari no script and random user agent for fingerprint avoiding. The OS is biggest give away though, apart from all location settings etc., disabling access to camera etc., there are still ways through, so then it is hardware and realising that the device is basically tagged. Tough for those who want a smartphone on continuously with personal details on it or using one with access to verified accounts (else you can do same as with dumb phone) , but otherwise it’s just an internet connection and computer like any other that can be traced, no big deal because “they know where you live” from a hundred other registries. So there are ways around it that aren’t too hard, but most of us don’t need to bother, I just learn it because it is something to know, plus I don’t like the idea of anyone aggregating data for whatever reason without permission.
Also, system details available from about any browser you use, just by using that browser, give a usually unique fingerprint to wherever you connect to that can be tracked. You can test and view those here.
At panopticlick you have to click
“Show full results for fingerprinting”
after it tests,for stats.
Random user agent confuses those, no script stops access to many details.
So what is a population of 8000 going to do with 40,000 lbs of butter ?
If they had any sense they would just buy a few cows.
That’s a years supply of butter for each person according to “average consumer data”. Maybe there is a pastry factory there…. maybe the truck got lost and was supposed to go to Mexico and they couldn’t get it to stop… and the residents of Quakertown are wondering why they are geting “free butter” offers with everything they buy!
And yet another sector of ‘higher middle-class’ jobs is about to go bye-bye…
Disruption Tax
There needs to be some kind of disruption tax so that the profits are not accumulated by the few, while the social costs are paid by the many….
I get shipping costs will drop, but the social costs (emotional as well as financial) will be significant. Truck drivers work hard and can make $100K/year. Now what?
Somehow we’re going to retrain them for another job? We’ve not shown any ability to do that so far….
I doubt it. That’s what lobbyists with huge bags of cash and prizes are for.
So true – it’s easier for you and I to pick-up the tab with two guys who designed all this getting super, super rich in the process and telling you and me that we’re supposed to be happy that shipping costs dropped a few cents……
There is a limit to economic efficiency and that limit is social stability….
The same arguments from buggy whip makers. Creative destruction is inevitable thats how free market capitalism works. Admittedly we have corrupt crony capitalism but still the same principles of economics apply to this scenario. Those jobs should be replaced by more productive and higher skilled jobs. Luddites have always argued this and so far there has never been the doomsday they predict. There may be some economic disruption but unemployment benefits and retraining is generally supplied by the govt. Even in the past without these benefits the workforce has adapted.
Actual social conditions at the time of the Luddites were pretty grim. A lot of death and destruction, so to speak. Literally. Starvation amidst plenty.
Primarily due to the Napoleonic Wars. Social conditions then sucked even so. Yet, the world survived when automated weaving machines replaced the skilled craftsmen doing it at the time. Most of this nonsense is chicken little stuff going on AI is not going to put everyone in the streets.
I want to be an optimist regarding “creative destruction” but how do you respond to the industrial revolution showing us the link is humans? In other words: When less and less were needed to farm so many more were free to provide time and labor to new technologies. So far, so good. However, these factories required HUMANS. Once AI/robotics is sufficiently capable of replicating the capacities of a trained and skilled worker in one field( with the attendant long run lower costs: no sick pay, works 24/7, no HR issues… ) why would those same technologies not be adopted across industries to achieve the same results? I’m not a Luddite, but our technological capacity is outstripping our ability to adapt as a society on an exponential level
You might be correct. However I have faith in human ingenuity and the ability to adapt. We are using and seeing things made possible through technology that I would never have even imagined even 20 years ago. Humans ability to capitalize on progress and new technologies has always resulted in an evolution of the work force. The question is will govt facilitate that bumpy road or make it worse like usual by encouraging people not to work at all.
Exactly…
“Hitler came to power on the back of extreme economic hardship and social unrest…”
the cause was as you ably outlined..
In Theory, practice and theory are the same thing. In practice, not so much…
The fact is that society can only handle so much change, so quickly.
The more rapidly people are displaced, the more likely civil order will be disrupted. Otherwise, why did the fed and the rest of the world’s central banks cushion the 2008 GFC? Answer: to avoid what would have been catastrophic social unrest…
Hitler came to power on the back of extreme economic hardship and social unrest…
That’s where we’re headed…
BTW Hitler came to power because of the economic disaster that resulted in unsustainable war reparations imposed by the allies after WW1. I think the world learned from that mistake also. To bad central banks can’t.
Exactly AND i’m getting sick and tired of this “higher skilled” argument. What? Truckers over the hill are going to back to school? That’s laughable! And no doubt those that embrace all this crap are probably anti-basic income as well. This is going to be a disaster because it’s going to be a rapid shift. They’ll see. And when the place looks like it’s burning down with continuous riots don’t say I didn’t say so. We are going to totally finish ourselves off with greed I have zero doubt.
Amen, brother! Amen!
I totally get that in the “long-term,” productivity improvements are critical, but there is a short and medium-term social cost that is ‘socialized’ among the masses…i.e., you and me…
The central banks made the situation worse. The book “The forgotten Depression” by James Grant outlines exactly what happens when the market clears leverage and failed business plans. In that instance there was a sharp and deep mini depression devoid of central bank f’ing things up. The leverage and flotsam was cleared and a more vibrant economy emerged within a year.
Actually come to think of it the best application for AI would be to replace the corrupt political class. They have no other useful skills and will then resort to street begging. Pretty much the same thing they are doing now, but for a little less renumeration. Perfect.