Self-driving trucks are more futuristic than the hype would indicate

The reality for driverless cars is that we’re at a ‘kindergarten-level’ with the technology it would take.

Missy Cummings doesn’t recommend choosing commercial pilot as a career.

This might seem like a strange statement for a pilot to make. Cummings admits to getting blow-back from her fellow pilots for saying so. Because before Missy Cummings earned her Ph.D. in systems engineering at the University of Virginia, she was a Navy jet fighter pilot. Currently, she’s directing the Humans and Autonomy Lab at Duke University.

“There’s a reason I chose academia over going into airline aviation,” she said. “Commercial airline piloting is mind-numbingly boring. Years ago, there were five people in the cockpit, and they were all necessary. Today, there are two people – soon to be one, and then maybe none.”

Cummings believes aviation will likely suffer the first real effects unmanned transportation will have on human workforce.

“Environmental ease is the advantage,” she said. “Today we rely on the pilot to make quarterback moves that require expertise, but almost everything in the process can be automated. We will see fully automated airline freight long before we’ll see driverless commercial trucks.”

Hype in marketing and media have increased a lot of the general public timeline expectation. Misunderstanding and incorrectly using terms associated with autonomous technology increases the hype. Headlines like “Trucking company uses autonomous technology to haul loads” can be extremely misleading.

The problem is “autonomous technology” can mean anything from cruise control to lane assist. An autonomous vehicle has any number of events in which it can do something without a prompt from the driver.

Pretty much every trucking company out there has some form of an autonomous truck. Autonomous features increase in increments to build upon multiple systems to maintain control without driver assist, or “self-driving” technology. But even “self-driving” is a misleading term.

Think of autopilot in a commercial airliner. Refer back to Cummings’ statement about relying on the pilot for quarterback moves, and note that autopilot is only applicable in some conditions. There is always a point at which control is expected to be relinquished to the pilot. This is also true with self-driving vehicles.

Driverless trucks are just that. There is never a point in which the passengers or freight are expected to do anything other than sit in the vehicle. That’s further in the future than a lot of marketing companies would like to promote.

“The reality is in the near future these technologies will get more advanced, but we are at a kindergarten-level for driverless cars,” Cummings said.

Research at the Humans and Autonomy Lab focuses on the multiple sociotechnological issues involved with interactions and integration of human beings with complex autonomous technology.

“We’d like to encourage the idea that this shouldn’t be a ‘war against the machine’ mindset but instead a partnership in which humans can, and will, be the coaches of this technology.”

Transitioning manual activities to fully autonomous is easier for some tasks than others. Cummings describes four levels of information-processing behaviors and the level of uncertainty involved in the activity that have to be taken into consideration with each level toward full autonomy.

“Skill” is a level in which an action itself is mastered. It’s much easier to automate a repetitive factorylike motion that never changes. There are no other factors to consider. Mastering a skill, such as drilling the same hole in the same place over and over again, is fairly simple, because there is little to no uncertainty.

Once skills are mastered, “rule-based,” or behaviors guided by subroutines, are considered. An example might be learning to turn on flashers while backing up a trailer, while mastering the skill of backing a trailer. There are two layers of processing behaviors in action. Multiple rules and multiple skills raise the level of uncertainty.

That’s where “knowledge” comes in. Practicing a combination of skill mastery and rule-routine fosters a knowledge-based level of function. Let’s take the trailer example a step further. A knowledge-based decision would be knowing that even if you can accomplish the skill and have your flashers on, you shouldn’t back over a human standing in the way. You have the ability to determine it would be fine to back over a plastic bag, but you know the difference between a plastic bag and a human being. The fact that your environment has such an enormous amount of variation increases the levels of uncertainty.

“Expertise” is the ability to combine all three levels of information-processing behaviors to successfully overcome extreme examples of environmental variation. Understanding a 48,000-pound load of bottled water will act a whole lot different in an emergency stop than a 22,000-pound load of toilet paper requires expertise.

The extremely diverse nature of commercial trucking requires a level of expertise that makes it an unlikely candidate for becoming fully automated anytime soon.

Cummings’ personal connection to the trucking industry allows her some insight to the specific needs and concerns of today’s drivers.

“My father drove a truck in the mid-’60s before he joined the military. My uncle drove a truck for as long as I can recall. I was raised in a very blue-collar family and understand the concerns drivers might have about automation.”

While Cummings said she believes that self-driving cars aren’t likely for at least another 20 years, she thinks trucking is a special case.

Fatigue is a serious concern among long-haul drivers. Some over-the-road routes that involve long, flat stretches of highway become monotonous.

“Health and well-being of drivers might improve if they were using driver-assist technology,” she said. “This won’t take their job away. It will improve their working conditions and make the job more appealing to younger generations as the technology develops.”

For the immediate future, trucking jobs are safe from total robot takeover. Driver interaction and equipment will change with the progression of autonomous technology, but according to Cummings, the time scale is probably longer than most people think.

“We have time to come together and ask questions about these issues, and the trucking professionals of today owe it to drivers of the future to start having these conversations.” LL