The Details of Work and the Realities of Automation

An interesting piece on the automation of trucking, with an extensive comment thread.  Many of the commenters have practical experience in the trucking industry and in automation work in other industries such as sawmills.

7 thoughts on “The Details of Work and the Realities of Automation”

  1. I drove truck for about 25 years. I owned a flat deck, then bought a tractor.

    A great deal of trucking is quite amenable to automation. It consists of warehouses filled and emptied, by largely, trucks. That is just a building full of storage space, often extensive shelving a lot of it computer tracked and some computer controlled. The warehouse has docks for the trucks to back up to, and really most of that could be quite automatic.

    Now flat deck trucking, moving objects that fit inside the legal space allowed, is a whole other thing. It involves a lot of judgement and some pretty off the wall solutions to sometimes unusual problems. I did a lot of that, it’s fun.

    It depends, is my take. ;)

  2. Given what Amazon does with their warehouses, where literally the one thing they use humans for is our hands to pick things up off of shelves, I don’t see anything that this article claims as being integral to trucking as not being already solved. Certainly figuring out how to pack a truck seems like something that a machine can do far, far better than any human could.

  3. The computations to optimize packing a truck are easily computerized, but the description of package contents needs to be available and interpretable in machine-readable form, and the physical manipulators (robots) need to be either available at the loading and unloading stations, or made a part of the truck (trailer, in most cases) itself.

    Not so difficult for high-volume shipments on fixed routes, much more difficult for the general case.

  4. I’m very much with Dan Hanson’s evaluation of driverless trucks.

    Few people appreciate the skilled manual labor involved at either end of a truck shipment nor the regulatory load on drivers between those two points of manual labor that haven’t been automated.

    >>computations to optimize packing a truck are easily computerized


    I spent 14 years auditing the load out of US Army 2.5 and 5-ton trucks for rail and road shipment. Accidents, lost loads of Army trucks and drug shipments put in Army trucks are all part of my work history.

    The biggest reason driverless trucks are not going to work is liability.

    Fired drivers solve most legal liability issues.

    Driverless trucks that crash with Hazmat shipments will see Google or Elon Musk’s company in court with their Indian programmer’s from Bangalore testifying about their code development issues for their truck to justify — and get — eight figure plus settlements.

    Plus the liability insurance for theft via hacking — as opposed to high jacking — loads will make sure -somebody- is in the cab of the truck.

    The liability insurance issue and the international air transport agreement (IATA) treaty are also reasons why the “pilotless Uber flying car” isn’t happening in the USA.

  5. This is an example of a 90/10 problem. Keeping the truck between the lines on the interstate seems simple enough and is where the most progress has been made. That is, until it snows or rains, road construction or something else. Even so, this is the easy part. The last 10% includes everything from loading/securing to maneuvering through side streets and crowded truck yards and will take easily an order of magnitude more effort. A solution that leaves out the last mile is worth a lot less than half of the whole thing.

  6. MCS,


    The 90/10 computer automating model is not a business model.

    The business model understand’s liability insurance is like the failed heat shield of a reentering space craft. If you lose the heat shield, you lose the mission. The same is true of liability insurance for on-going businesses.

    In terms of trucking firms, it is a great thing to have a driverless rig on know routes with material handling services on both ends. The reduced manning and healthcare get turned into a depreciating capital asset.

    The same is manifestly untrue for the manufacturers of driverless big rigs.

    Driverless big rigs set up Google and Tesela as the “Deep Pockets” for hundreds to thousands of auto versus semi-tractor trailer accidents a year.

    Ditto things like smuggling in unmanned vehicles and hacking of same to cause accidents or other criminal acts.

    IOW, driverless vehicles are a form of viral transmission of legal liability from drivers to deep pocket manufacturers. A viral transmission that insurers cannot calculate the risk for to properly charge Google et al with rates that cover the Insurers losses with a reasonable profit.

    If the price of using driveless vehicles plus unknowable risk insurance is higher than manned vehicles in the same application. You are going to see manned vehicles in the application.

    There will have to be a couple of decades of military – government history operating “Crew optional” military logistics vehicles before you will see any mass take over of trucking by Google/Tesela semi-rigs.

    This is a classic example of the leadership problems involving the “Credentialed but uneducated” and “don’t understand cause and effect” SJW elites now running tech companies.

  7. This Instapundit piece is what the SJW leaders in Tech need to read and heed about the obstacles to a transition to driverless trucking, but won’t —

    RISE OF THE MACHINES: Virginia Postrel: Lessons From a Slow-Motion Robot Takeover: Cotton harvesting is now dominated by machines. But it took decades to happen.

    The story of how cotton harvesting has changed over the decades doubles as a reminder that even robots take their time. At least until a certain point.

    1) Full automation was impossible without years of tinkering. Although mechanized cotton harvesters were available in the 1920s, they didn’t catch on until after World War II. As long as farms needed workers to hoe weeds and thin cotton plants, replacing them at harvest time made little economic sense. Chemicals, not machines, solved that part of the problem; the ground between rows in Terry’s field is perfectly bare.

    Even that wasn’t the end of it. “The ancillary requirements seemed to go on and on,” wrote the late historian Donald Holley in The Second Great Emancipation: The Mechanical Cotton Picker, Black Migration, and How They Shaped the South. Gins had to install dryers, for instance, because machine-harvested cotton retained more moisture. Farmers needed chemical defoliants to apply before harvesting so that their bales wouldn’t be contaminated with leaf trash. Breeders had to develop shorter plants with bolls that emerged at the same time, allowing a single pass through the fields. Until all these things had happened, harvesters had limited appeal.

    Replacing human adaptability and skill, in short, required much more than a single new machine. Production systems are far more complicated than outside commenters realize. Robots may eventually replace people in an industry, but it can take a long time.

    Read the whole, very interesting, thing. And somebody tell George W. Bush that we don’t need immigrants to pick cotton any more.

    85 Posted at 10:30 am by Glenn Reynolds

Comments are closed.