Hacker News Reader

Comment by hughes | Hacker News Reader

Maybe transport & trucking needs to be automated. If there's a lack of labourers, why not fill in the gaps with self-driving semis?

View on HN

Not surprisingly this is the first step in that process. Automating the driving of trucks is expensive, there was at one time an ample supply of truck drivers keeping wages low, so no automation. As the cost of human drivers goes up, it changes the math with respect to automation, once it hits the tipping point the change over is inevitable.

We saw an exact corollary of this with automation in Chinese factories. As long as the cost of labor was lower than the cost of automation factories remained manual only shops, but the cost of labor has gone up, and with it have come robots. What I find particularly fascinating is that in the labor case the manufacturing went off shore first (it was cheaper to hire Chinese workers) and now it is being automated, in China. Since you cannot easily off shore truck driving that option doesn't exist.

I do however keep my eye out for 'teleoperated' trucks. Something which is a mix of robotic driving with supervision and command and control with a human. This is also possible today and would allow a trucking company to operate more trucks with fewer drivers, but it also requires some automation on refueling. The system has resisted this form for a while.

So what I took away from the NYT article was that the economics of trucking are entering the point where the system can support more dramatic changes than it has in the past. That means there are opportunities there.

I think we're quite a bit farther away from self-driving semis than from self-driving cars. Big trucks are much more dangerous vehicles than cars: their braking distances are much longer, their weight makes it possible for them to destroy any cars in their path, they can require multiple lanes to make a turn, many car drivers don't know how to drive safely near trucks, etc.

No kidding. I'll get downvoted for this but the Hacker News demographic hasn't spent much time on the road. Try driving down 710 toward Long Beach. Last time I lived down there, tons of trucks, tiny lanes, uneven pavement, completely insane. I want to see a self-driving car navigate Manhattan or Brooklyn during the day, normal driving rules do not apply there, New York drivers mean business ;-)

Sure, but the entire 710 just needs to be demolished. They should be moving those containers out of the port on trains, but they've got it backwards. They move the containers out of the port on trucks, to intermodal terminals inland where they get switched to trains. Completely backwards. If they just punched the railroads through all the way to Long Beach then they wouldn't need those trucks on 710 humping boxes to the Inland Empire.

I haven't been there in over 10 years but it was the most direct way by car to get to Long Beach from say downtown Los Angeles. It was strange to see that industrial wasteland along 710. I tried to find a shortcut through there once and ended up on a distant, post-apocalyptic planet ;-)

> If they just punched the railroads through

Isn't that what the Alameda Corridor project did? https://en.wikipedia.org/wiki/Alameda_Corridor

Yes, but they still don't have the capacity commensurate with the number of boats offloading at the ports.

The promise of self-driving cars is that they will be better and safer drivers than humans. Wouldn't that imply that they would be even now suitable for high-skill, dangerous driving?

That's "the promise", yes, but delivering stuff is very different to promising stuff.

Planes are way ahead of trucks in this regard and could actually fly themselves from origin to destination most of the time, but we still keep a couple of highly-trained pilots sitting there to handle the rest of the cases.

A truck can kill nearly as many people as a plane, depending on its cargo and just how badly it screws up. In more common cases, though, I'm unconvinced by software's ability to figure out the best thing to do when it's travelling down the highway at 100 km/h and it sees something unexpected dash onto the road, with a better error rate than a human brain, which can at least immediately distinguish between humans and deer.

Computers can distinguish between human and deer with 100% accuracy in microseconds. Human brains aren't even close. You might as well compare a sprinter to an F1 car.

Theoretically seems easy but I think practically it's incredibly difficult due to the unpredictable nature of Over the road trucking.

There's no automated cross country trains and those move on a predictable path.

Although I would love to see it happen because of how incredibly efficient it would make the transportation industry.

Long-distance trucking seems like a huge and obvious market for autonomous vehicles. Trucks mostly travel along Interstate highways and drivers have mandatory rest periods. Usually that means the truck sits somewhere, sometimes with the engine on.

As far as I understand, many places use trains for these purposes. It means that less drivers are needed and the cargo spends more time in transit.

Alas, the united states has a fairly poor cargo rail system, which in fairness is due to its spread out geography in many places.

I don't know that much about the railroads but that doesn't sound right:

"America's freight railways are one of the unsung transport successes of the past 30 years. They are universally recognised in the industry as the best in the world."

http://www.economist.com/node/16636101

No, in the US the two systems are used for different things. Trucks allow point-to-point delivery. Trains are more useful for carrying bulk commodities.

John McPhee has written some wonderful articles about modes of transport, and the people who work in them: http://www.amazon.com/Uncommon-Carriers-John-McPhee/dp/08654...

I was about to post about John McPhee, too, when I saw your comment. Excellent book, recommended.

Somewhat related, also great: Looking for a Ship.

The Box, but Mark Levinson is also a great read. http://www.amazon.com/The-Box-Shipping-Container-Smaller/dp/...

What's stopping intermodal containers being used to displace the long-distance trucking? Trains should on the face of it be cheaper, even when you're not transporting bulk commodities (you're talking in the US case a train 1500m long, potentially with the containers double-stacked depending on the route).

This paper seems to give an interesting perspective: http://intermodal.transportation.org/Documents/brown.pdf

It comes down to highway transportation being massively subsidized in the US, to the detriment of rail. So rail is cheaper but much higher latency than trucking, with the potential for long and unpredictable delays.

Most trucking companies are prohibiting idle running, they program their trucks to shut down after a given time if the truck is not moving.

Most rest stops have also started installing "IdleAir service" which supplies Heat, AC, 120V electric and in some cases Cable TV, at the drivers expense of course.

Reply on news.ycombinator.com