Who will clean self-driving vehicles?
I found myself wondering this recently as my son and I tidied the family car after a road trip. We’d been driving for only five hours, but we had produced two grocery bags of trash: water bottles, parking stubs, wrappers from lunchtime hoagies, reading material, a roll of Scotch tape, and a ping-pong ball among other miscellany that had accumulated over the short time. It wasn’t unusual. In my family, I’m the one who remembers to clean out the car, so I’m all too familiar with the volume and medley of mess that can be generated in vehicle regularly used by adults and kids.
Yet with companies like Uber, Waymo, and Lyft planning to launch their first generation of self-driving cars as shared taxis, it’s not yet clear who or what will be there to clean up the half-drunk Starbucks cup, wipe down the mystery stickiness on the seat, or handle even less hygienic situations. It’s not just a trivial matter: it’s an issue of sanitation and rider well-being—one more pressing for future users than you might imagine.
Consider the many dimensions of mess. As I thought about mess in cars, I wasn’t just thinking about cleaning up the slightly gross piece of lettuce from my son’s hoagie that had fallen on the floor mats. I was thinking about cleaning up an even grosser kind of mess—the kind that you make if you are carsick.
I have to admit something embarrassing at this point: I am hideously prone to motion sickness. Cars. Boats. Planes. Subways. All of them make me nauseous. Sometimes, the consequences are messy. Reading of any kind makes the problem worse. This includes interacting with screens on mobile devices, which, unfortunately for me, a Carnegie Mellon survey suggests will likely become the most popular activity for self-driving car riders.
I started asking around to find out if other people had thought about motion sickness and self-driving cars. I found University of Michigan researchers Michael Sivak and Brandon Schoettle, two scholars in the vanguard of thinking about robot cars and retching. In a 2015 report, “Motion Sickness in Self-Driving Vehicles,” they concluded that people riding in self-driving cars are more likely to experience motion sickness than people riding in conventional cars. Since motion sickness most often occurs when the movement you feel doesn’t align with the movement you see or otherwise anticipate, the pair saw multiple potential triggers for the ailment more likely to be at play in self-driving cars. Among them: a lack of control over direction and speed of movement, a non-forward-looking gaze, and an increased likelihood of participating in activities like reading, texting, watching videos, or playing games. Because of this, according to the researchers’ calculations, 6 percent to 10 percent of American adults riding in self-driving cars are likely to experience motion sickness often, usually, or always.
Think about that for a moment: At least 1 out of 10 self-driving cars will often contain someone who is experiencing motion sickness. That’s going to mean an awful lot of cleaning up.
Notice that this estimate doesn’t include children or mention pregnant women. Yet they are the first groups that I, and probably most of you, think of when considering who’s most prone to the Technicolor yawns. Most families have some kind of gross story about their kids traveling and tossing cookies. “My son Owen has hurled in the exact same mile marker range on the PA Turnpike, between Mountains and Breezewood, on every road trip to Indiana,” writer Monica Yant Kinney told me about her 10-year-old son. “That’s two pukes per trip, three to four trips per year. He is a pro at using a gallon-sized Ziploc bag. No mess!” Some research also suggests that women may be more prone to motion sickness than men.
It’s already common for passengers to ralph in ride-share vehicles. Typical is the story a graduate student recently told me about a time she ordered a Lyft after an 11-hour flight from Tokyo to Boston. “I’m prone to motion sickness, especially around menstrual stuff,” she said. After the long, exhausting flight, she got into the car with and started chatting with the driver. Then it hit. “All of a sudden, I smelled a floral scent from somewhere in the car, and I started feeling nauseous,” she said. The driver gave her a full bottle of water, which, she said, turned out to be the wrong strategy: “I took a sip of water every time I felt nauseous. Then, I projectile-vomited on myself and the car.” They pulled over, at which point the student opened the door to finish outside.
She said the kindness of the driver—stopping the car, asking her if she needed a minute to breathe, giving her a box of tissues, showing that he wasn’t angry with her—meant a lot in what was an uncomfortable and embarrassing situation. Still, it was all over her and all over the car. She said she apologized profusely. She also said she tipped him $25 on a $25 ride, and thus avoided the $25 to $250 fines that Lyft, Uber, and some municipal taxis are known to allow drivers to collect for such “damages.” After all, it’s on the driver to clean up the mess.
Motion sickness is only one dimension of the many reasons that people yak in cars. There are also the drunk people, and the babies who spit up, and the people who are ill with gastrointestinal issues. There’s other unexpected discharge, too. One woman recently wrote in the Guardian about giving birth in an Uber: It was, like all births, messy. Sick and injured people are also known to take taxis or rideshares to the hospital instead of calling a pricey ambulance. But average cars, including the driverless models currently in testing, do not have the wipe-clean surfaces of an emergency vehicle’s interior, let alone the plastic-y seat covers of many cabs. Have you ever wondered why New York City taxis have smooth vinyl surfaces instead of molded plush seats of conventional passenger cars? It’s a design decision that comes out of long experience with a wide variety of humans.
Then there are the other body fluids—the ones usually managed in private. In Driverless: Intelligent Cars and the Road Ahead, authors Hod Lipson and Melba Kurman write about autonomous vehicles’ potential to become a kind of rolling No-Tell Motel. In one scenario, they imagine a new line of driverless cars that includes a “bed bus” model, complete with shaded windows for privacy. In other, they imagine a comfortable mobile environment for viewing entertainment on screens or with virtual reality goggles. Naturally, this could include one particular type of media product: porn.
This notion that driverless cars could become pornography hubs on wheels is quite common in futurist circles. Jeffrey Tumlin, a strategist at transportation consultancy Nelson/Nygaard and interim head of the Oakland, California department of transportation, predicts that self-driving cars could become vehicles for prostitution services—essentially fitted out to be self-driving brothels. Putting aside for a moment the very serious questions a sex-on-demand robot car raises about safety and human trafficking, think about the mess. Perhaps these vice cars will specialize. But even so, it’s still not a stretch to imagine that people—teenagers, adulterers, or other couples caught up in the moment—will avail themselves of the perceived privacy of a driverless car during a standard trip. Do you really want to be picked up by a self-driving taxi that has someone’s used condom on the floor?
With all the mess—organic or not—with the potential to build up in shared self-driving cars, it’s hard for me to accept the sleek future many envision for shared autonomous vehicles. Instead, lacking the human drivers, I see self-driving taxis going the way of the Port Authority Bus Terminal bathrooms. Circa 1980. Before they had bathroom attendants.
Some tech has been proposed to deal with preventing some of the mess, including motion sickness. In 2016, Uber filed a patent for a “sensory stimulation system for an autonomous vehicle” designed to combat it with haptic feedback (think lights, vibrations, or bursts of air). Waymo also has patents for devices to curtail motion sickness, including a head-mounted VR device. As someone who’s experienced motion sickness with every VR device I’ve ever tried, I was more optimistic about the Uber strategy.
I spoke with Molly Nix, the UX lead for self-driving Uber cars, and one of only two product designers working on what the company deems the “self-driving Uber human experience,” which includes everything from the app interface to the logistics of motion sickness. As it turns out, Uber’s haptic feedback technology might not become reality. Nix explained that the patent is a reflection of the kinds of things the Uber team is thinking about, but that, “It’s important to remember there is such a thing as overengineering a solution to a problem like motion sickness,” she said. “Nothing beats windows.” Staring outside may be the best remedy for passengers, and choosing when you need to open a window may be better than relying on a hyperdesigned haptic feedback system giving you bursts of air.
But even less thought seems to have been put into cleaning. When I asked Nix what would happen if someone made a call on a porcelain telephone in a self-driving car, she declined to answer. I asked if she and her team talk about it at the office. She again declined to answer. What will any kind of self-driving car garbage cleanup look like in reality? “We are still envisioning what it might look like,” said Nix.
Sarah Abboud, an Uber spokesperson, said that the company doesn’t have a plan for dealing with the aftermath of people getting sick or making other serious messes in self-driving cars, in part because the vehicles Uber’s testing now still have backup human drivers. “Since we have an operator in the car, we have not really explored exactly what that looks like,” Abboud said. She added she imagines that such messes would probably be handled in the same way the company plans to handle general cleaning: dispatching the car to a facility for a human to clean it and get it back on the road. There are currently two operation centers that clean the driverless cars Uber is testing, one in Phoenix and one outside Pittsburgh. Perhaps Uber would create more of those, Abboud suggested.
The same seems to hold for other companies. Waymo, for example, has partnered with rental car titan Avis for routine maintenance of its self-driving vehicles in Phoenix—though the few available details a Waymo spokesperson sent to me simply suggest that cars will “need to be charged and refueled, cleaned, and presentable for riders.” The overview did not include information about how, exactly, this happens. (Lyft did not respond to a request for comment on the cleaning issue.)
It’s possible that companies could program cars to return to a home base for upkeep after every ride. But it’s an unlikely solution considering the potential for wasted time, wasted energy, and increased congestion. Instead, as of now, solutions still seem to rely on human intervention. Someone will likely need to alert Uber or Waymo to any mess in a car. Then someone will need to clean it. (No Roombas for car interiors yet.) Abboud alluded to a potential mechanism that might help Uber’s systems identify such messes in the future, but wouldn’t say if that would be a video camera inside the car or something else. “We don’t really have that figured out yet,” she said.
The lack of strategy from these companies seems to reflect a naïve view of what people are like and of how much invisible labor drivers put in to keep their cars clean. According to Nix, Uber expects passengers to take out their own trash today, and “expect[s] that to continue for self-driving ride-sharing.” But that ignores the way the dynamics will inevitably change. Today, people may have social guilt about intentionally littering in the car of a person inches in front of them in the driver’s seat. Even if they don’t, if a passenger leaves the odorous, crumpled remains of a late-night McDonald’s run or a mess of a more biological nature behind in a taxi, a human driver will be there to see it, clean it up, and maybe roll down the windows to air out the car. Oh, and to ding the passenger with a cleanup fee. A human can do these things easily. A robot can’t: There’s no sensor for grime, mess, or stink.
When the driver disappears, we already know what happens: Cars get dirty, smelly, and damaged. Just look at Zipcar, the car-sharing service owned by Avis, which relies on an honor system for cleaning. As Business Insider detailed, Zipcar’s currently got a “D-“ rating from the Better Business Bureau for customer service—and has been struggling for it—in large part because of complaints from users who find their rented cars are dirty and damaged from past use. Sure, customers can report such issues to Zipcar. And, in the future, they may be able to use an app to flag unhygienic cars for Uber or Waymo. But that’s a lot of time and frustration to potentially put on users.
I’m not surprised that companies like Uber don’t have a robust plan to deal with cleanliness. The company seems to exemplify the kind of biased worldview that I call technochauvinism. Technochauvinists tend to prioritize technical issues solved by engineering and math, while overlooking the human factors that shape how platforms or systems can and are used in practice. This results in strategies like pouring millions into developing apps that edge out the competition and computers that get its driverless cars onto the road fast over making sure that, for example, women passengers don’t get harassed or assaulted, or that people like me don’t get physically ill from riding in its cars. However, if you are trying to design a fully autonomous system that involves 2-ton potential killing machines—which is what robot cars are—the stakes are high. Human factors matter a great deal, and not just because humans are the other drivers on the road.
Fortunately, I can also attest that there is one surefire cure that works every time I get motion-sick as a passenger in a car. Instead of riding, I drive.