In 2016, Uber delivered 50,000 cans of beer over 120 miles of highway without any driver intervention. But two years later, Uber closed down its self-driving semi-trucks division to focus on self-driving cars instead.
The trucking profession is huge, over five percent of all American jobs are in truck transportation. When self-driving trucks progress far enough, well over a million Americans very well may find themselves out of jobs. With all the clamor regarding autonomous vehicles, why is it taking so long for them to get here?
In 2017, Susan Carpenter of Trucks.com cited “industry experts” that autonomous big rigs would be much less of an anomaly by 2020 or 2021. As of writing this article, self-driving semi-trucks are so rare that them making deliveries are still newsworthy.
Why has the technology failed to match even a fraction of the expectations industry experts have set? This might seem odd, but if one were to look at past inventions’ market penetration, it is not very absurd.
Take a look at the adoption history of cars in general. The invention of the gasoline powered car was in 1895, but it took roughly fifty years until 60 percent of American households owned a car, and another sixty or so for 95 percent household penetration.
Autonomous vehicles have only recently started becoming a reality. To say that they would rapidly replace cars and semi-trucks as the new standard was a pipe dream.
Man Versus Machine
If every vehicle on the road were replaced with fully autonomous (also known as level 5 automation) vehicles tomorrow, it would overcome the human element. Human drivers are one of the single biggest hurdles self-driving vehicles must overcome.
Humans are hard to predict. A self-driving truck could wirelessly communicate with other autonomous vehicles and make adjustments by the second to accommodate.
A self-driving car currently has no method to communicate its plans to a human, aside from the common turn signals and headlights. Because most drivers are worse at driving than they like to think, a self-driving truck could predict what another driver does incorrectly, leading to hard braking that damages cargo, or worse, a collision.
The opening example of Uber having a semi-truck deliver 50,000 cans of beer 120 miles has a glowing caveat: those 120 miles were all done on a highway. To get the truck to and from the highway, a human needed to be in full control of the vehicle, who then hit the autopilot button once the vehicle was on the open road.
Currently, the technology for the smartest driverless cars and trucks is at level 4 automation. This means that the driving computer can handle most cases, but cannot account for uncommon road hazards.
Roads that are not highways and freeways have a much greater scope of hazards. There are dozens of scenarios such as pedestrians and school zones that require a lot more driver adaptability than an autonomous vehicle currently can be programmed to have. As such, trained humans will still have to be in the cabin to be able to take over for manual override.
Truckers might very well be paid less per mile because they will spend less time actively being in control, but it will be a very long time before they are up to the task of handling every individual circumstance like humans can do.
Susceptible to Hacking
The automated vehicles’ inability to adapt can be highlighted most strongly by a “hack” that requires no programming knowledge. Merely the act of placing stickers on street signs can trick machines into thinking things that no human would ever consider, such as a stop sign merely being a form of street art.
Driverless vehicles are computers, and like most computers, they can be hacked to give access to malicious actors. Companies behind the autonomy revolution are on the case to make hacking these cars as difficult as possible through methods such as having multiple sensors that would need to be fooled in order for the car to drive off course.
Because hackers can use bugs in the system to make a semi-truck think speed limit sign is 30 MPH higher than it actually is, having humans be the primary operator while the computer is an assistance system is currently the optimal use for the technology. As the vehicles receive software updates to better understand signs, the reliance on a human agent may very well go down eventually.
In a normal accident, there tends to be some clarity as to who is to blame. A vehicle having autonomy has caused a question to arise as to where owner responsibility ends and where manufacturer responsibility begins. It is common for humans and companies alike to minimize their exposure to any chance of an overwhelming lawsuit.
Tesla is a shining example of this: in their cars with enhanced autopilot, the driver who wants to use it must follow a strict set of guidelines. They must agree to a waiver every time they want to turn it on, and must keep their hands on the steering wheel at all times, ready to take over at any time. Finally, the system is only usable on freeways and highways, automatic driving on city streets is prohibited.
Now, imagine the risk involved with a semi-truck carrying hazardous materials such as dynamite or sulfuric acid. Needless to say, there is a lot of ironing out when it comes to liability rules before automatic semi-trucks become a mainstay on the road.
Current truckers do not need to perform a job search; trucking is here to stay. While self-driving vehicles are less likely to get into an accident on open, straight freeways, they currently lack the technology for level 5 automation. Other humans and extreme circumstances such as flash flooding and icy roads make a human’s ability to rapidly adapt to changing situations too valuable of an asset for the transportation industry.
If the current state of autonomous vehicle technology disrupts the trucking industry in any way, it will be trucker pay. The driver of an automatic truck will spend significantly less time paying active attention to the road, and the pay will be updated to reflect that. For some this will be a worthy trade-off, as truckers get paid a third of the rate to put in a quarter of the effort, but for others this reduction in wages can very well price them out of the market.
It is also possible that there may come a new form of Commercial Driver’s License that only grants driving semi-trucks for scenarios that a truck’s computer otherwise cannot handle. This would increase the supply of available workers for these trucks, also pushing wages down.
If self-driving vehicles adoption parallels standard vehicles in its history, it may very well be decades, not years, before autonomous commercial vehicles become the new standard.
Only time will tell when autonomous driving becomes a standard for semi-trucks, but with all the obstacles the technology still has to overcome, it will be an absolute miracle if those industry expert predictions from 2017 become true.