OneZero, Dave Gershgorn, November 24, 2020
‘If the truck I’m in ever gets a camera installed in it facing me… I will stop the truck and quit on the spot’
Over the last three years, hundreds of thousands of freight trucks in the United States have been equipped with machine learning algorithms to analyze drivers’ behavior. They can detect how many times per trip drivers pick up their cellphone, get distracted while driving, or even just look fatigued when they’re behind the wheel.
The tech is built into driver-facing dash cameras, which have been adopted by the trucking industry over the last 10 years. These cameras have already been a contentious issue in the trucking community. But the ability to proactively recognize behaviors in the cabin using machine learning adds a new layer of surveillance to the cameras. Rather than just reacting to a hard brake or a swerve of the truck, the A.I.-powered cameras respond to the driver themselves.
Robot Truckers Are Here — They Just Happen to Be Human
Electronic tracking devices monitor just about everything human drivers do, automating what was once an independent job
Dash camera companies say that their technology makes the 40-ton trucks barreling down our roadways safer, and only alert fleet safety managers to dangerous driver behavior. In fact, companies claim, the cameras protect drivers, too, since footage can exonerate drivers from “nuclear verdicts,” an industry term for lawsuits against drivers involved in accidents with payouts over $10 million. Insurance agencies will also lower rates for fleets with driver-facing cameras.
But the Teamsters, America’s largest trucking union, say that the technology leads to micromanaging and is an invasion of members’ privacy.
In 2017, truck drivers for Sysco in Quebec won a five-year lawsuit over driver-facing cameras, claiming that the cameras sometimes recorded for no reason; one driver even said he felt distracted by constantly checking to see if he was being recorded or not. The Quebec court eventually ordered cameras to be removed from Sysco trucks in favor of “less intrusive” tracking methods, according to Truck News.
“If the truck I’m in ever gets a camera installed in it facing me, facing inside this truck, this home where I live in, I will stop the truck and quit on the spot,” said YouTuber Trucker Josh, a trucking influencer with more than 100,000 subscribers, in a video posted in 2017.
The conflict pits the autonomy and privacy of truck drivers against the potential safety of the drivers and others on the road.
Most A.I.-powered driver-facing cameras function the same way. The camera, which is connected to the truck’s internal computer, constantly records footage of the driver in the cabin when the truck is on. This footage is only saved and transmitted to a manager when the camera detects some kind of risky driving.
The conflict pits the autonomy and privacy of truck drivers against the potential safety of the drivers and others on the road.
Many of the biggest names in the dash camera industry now foreground “smart” capabilities. Firms like Lytx, Samsara, and KeepTruckin have started to add algorithms to their hardware in the last three years.
Companies like Lytx, which claims to have 650,000 cameras deployed in commercial trucks across the world and 60% market share, say that people are alive today because their technology stopped crashes. Lytx’s camera has a small computer onboard that runs machine learning algorithms to detect around 60 behaviors that are determined to be risky. These behaviors include driving without a seat belt, looking at a smartphone, or smoking while driving.
If a fleet manager wants to see the footage, it is uploaded to the cloud to be analyzed again, and then to a human reviewer employed by Lytx. After it’s been manually reviewed, the incident and footage appears on the Lytx dashboard for fleet managers to track their drivers.
Lytx’s cameras try to detect the driver’s behaviors and alert the driver before sending anything to a manager, which Nixon says gives the driver an opportunity to correct their behavior.
“We can now give alerts to the driver to help them self-correct. We can tell them the percentage of their trips that they have picked up a cellphone. We can tell them the percentage of time that they’re driving where they’re distracted or fatigued. And it’s all about having the driver self-correct when they realize this information,” Lytx CEO Brandon Nixon told OneZero.
Instead of monitoring drivers’ behaviors to track unsafe driving, a company called KeepTruckin uses driving data from the truck’s computer to determine when the camera should record, as well as algorithms to scan the driver-facing video. If the truck brakes heavily or swerves in a pattern associated with distracted driving, the camera will activate and log the behavior. The company says A.I. also triggers the video to record, like if the driver is tailgating or distracted.
These behaviors include driving without a seat belt, looking at a smartphone, or smoking while driving.
KeepTruckin’s dashcam business started with outward-facing cameras, according to Dhruv Maheshwari, the company’s product manager for A.I. and dashcams. But truck fleet managers told the company cameras that face the road don’t tell the whole picture.
“Fleet managers wanted to start getting a better understanding of what the driver behavior actually was, and are there behaviors that could be improved,” Maheshwari said.
Samsara uses machine learning algorithms to analyze the footage in real time and alert drivers to unsafe driving. A video of the company’s technology shows how the movement of a drivers’ head is tracked in real time as they look down at their phone, causing an alarm to sound. A company spokesperson tells OneZero that to protect driver privacy, they only store footage when distracted driving or a “harsh event,” such as an accident or fast braking, has occurred. Samsara also provides drivers with a lens cap to cover the camera, and records which manager or supervisor have viewed a recording. The company claims to have saved its clients an estimated $225 million in legal fees by using video to defend drivers.
The Teamsters and some advocates argue that no matter the privacy precautions, having a camera pointed at you while working invades privacy and adds to the stress of the job.
“From a driver perspective, there have been studies that show in other industries that workers who were surveilled electronically experience increased job stress,” says Lamont Byrd, director of Teamsters Safety and Health Department. “Especially in an industry like trucking, where you’re already having to deal with weather conditions, traffic, tight delivery schedules, long work hours, and now in a pandemic, it’s just something that’s added to a list of challenges that they’re already facing.”
Byrd contends that with access to the truck’s computers and outward-facing cameras, fleet owners should have more than enough information about how a driver is working and to log accidents.
“There’s a way to get at this data, in terms of what actually happened and the driver’s response, without constantly monitoring what a driver is doing,” he said.
