With so much attention being paid to the development of self-driving cars (including whether Apple is getting into the game), little notice has been given to the prospect of self-driving leviathans of the highway—the long-haul truck. Industry watchers and tech fanboys are salivating at the potential of making their commutes or jumping between meetings in autonomous people-movers, yet, in terms of near-term impact, the self-driving semi could make a bigger splash given the trucking industry contributed $642 billion to the US economy last year.
Take the initial significance that self-driving trucks might have on the road: Based on the most recent data (pdf) from the Department of Transportation, there were an estimated 10.7 million trucks operating in the US in 2013, many of which belong to the more than 33,000 commercial fleets on American highways. While enticing individual owners of private vehicles to put self-driving cars in their driveways could take decades and only change one car at a time, truck fleet owners could replace potentially dozens of vehicles in a single stroke, particularly considering the potential of economic incentives and regulatory mandates. Transport experts promise economic benefits in the form of reduced fuel consumption and emissions through more efficient routing and smoother driving, longer routes than people can safely drive, and less space taken from private drivers on the road. Trucks also have to deal with fewer stop-start, frequent turn scenarios, and fewer road obstacles, meaning a possible less daunting learning curve for self-driving systems.
Google roads
Dr. Anthony Townsend, senior research scientist at NYU’s Rudin Center for Transportation Policy & Management envisioned one of four scenarios in the future of transportation in America called “Re-programming Mobility” (disclosure: I was a pre-publication reviewer). The study, which imagines distinctly different ways that our lives, cities, and movement patterns might change focuses on outcomes driven by automation: everything from co-working jitneys to city streets handed over to delivery drones at night.
In the study’s “Growth” scenario, Townsend sees a push for self-driving trucks partly as a result of converging economic, technological, and regulatory factors. In this world, by the late twenty
-teens, Google proposes partnering with the federal government to overhaul the road system, creating “G-Roads.” This massive investment, made for explicitly economic reasons, trickles down to self-driving cars, laying down the technology consumers would need to make owning a private self-driving car more valuable.
Writing from the perspective of 2020, Townsend describes the evolution that takes place in this future. “The first step [in rebuilding a stagnant economy] was a crash program to automate long-haul trucking,” he writes. “With the driver shortage that had grown throughout the 2010s, and the continued expansion of e-commerce, trucking companies were increasingly struggling to keep goods moving. Citing national security concerns, the federal government established an aggressive timetable for full conversion. Freight haulers were only too happy to comply, and tens of thousands of vehicles were upgraded, with significant cost savings, emissions reductions, and most importantly—fewer accidents.”
However, in the study’s “Collapse” scenario, this collaboration is resisted by unions and car manufacturers not keen on having Google gain the same leverage over mobility as it has today over internet search. In this scenario, resistance to such a big shift looks more like a slowly emerging French-style blockade, gumming up roads of 2020 America in protest of poor planning. As a result of this slowdown, “there were more trucks on the roads than ever, and they were a major disruptor to the formation of computer-controlled platoons that could speed the flow of traffic,” writes Townsend.
Early signals, uncanny outcomes
Of course, self-driving trucks aren’t just the stuff of future scenarios. Pieces of the puzzle are already taking shape. Mercedes is already showing off demos of self-driving trucks. A company appropriately called Peloton has been working on the capability of platooning autonomous trucks to gain greater economic efficiencies. And of course, the omnipresent Elon Musk has proposed even skipping highways altogether and putting road freight on his Hyperloop tube transport system.
Major delivery companies such as Fedex, UPS, and DHL have been refining the algorithms that direct their human drivers’ routes for years though this is now reaching the point where the drivers and the algorithms are seeing eye to eye less frequently, and drivers’ learned experience from years on the road is clashing with the cold logic of software, where routes can seem less intuitive to a single driver even while being more economical for the whole company. DHL, for one, sees self-driving trucks as completing an already rapidly automating supply chain, where container ports on one end and distribution warehouses further down the line are already filling with robotics.
Unemployment or upgrade?
Given that the most recent US Census lists “truck driver” as the most commonly held job in 26 out of 50 US states, the potential for employment disruption would seem substantial. As has been indicated by Uber’s recent announcement that it is researching self-driving cars, one next step for software that routes human drivers is to use this knowledge to skip the chauffeur altogether and let canned intelligence take over. Economists are starting to warn of this broader pattern of jobs “below the API,” or jobs that can be understood as a basic set of actions which can then be coded in software, as being under threat. Driving a truck would seem to fall into this category.
Given that the most recent US Census lists “truck driver” as the most commonly held job in 26 out of 50 US states, the potential for employment disruption would seem substantial. As has been indicated by Uber’s recent announcement that it is researching self-driving cars, one next step for software that routes human drivers is to use this knowledge to skip the chauffeur altogether and let canned intelligence take over. Economists are starting to warn of this broader pattern of jobs “below the API,” or jobs that can be understood as a basic set of actions which can then be coded in software, as being under threat. Driving a truck would seem to fall into this category.
Yet truckers who are good at their jobs might actually see a move to the back office—of the truck itself. In this scenario, self-driving trucks would remain staffed with something more akin to a human first officer, someone sitting behind the cab in a vehicle, perhaps part of a platoon of cooperating autonomous trucks, managing routing, interacting with customers face-to-face and taking over the wheel in situations that require more sophisticated action—in effect herding a 10-ton “just smart enough” beast of burden. The dispatcher, not the driver, really may be the one under threat.
The deeper future gets a bit more weird. One suggestion DHL has considered includes delivery trucks that follow human delivery staff on foot down a street at a distance, like a robotic pack mule, bringing to mind visions of Boston Dynamics’ robotic Big Dog, clunking down the road, bearing grandma’s Christmas cake to your door. The upside would be keeping the human delivery touch—where customer knowledge still counts (instead of said Big Dog tossing a package onto your veranda). Of course, as proposed by one ex-Google engineer, that pack mule might just be a subcontractor, owning itself and working its own routes like an AI-driven TaskRabbit working for Bitcoin.
Beyond that, well, it’s all space truckers.