Texas A&M University is modifying its self-driving pilot program in the city of Bryan, Texas, to have humans remotely monitor and operate the shuttles starting in September, making it one of the first commercial deployments of teleoperation technology in the country.
The teleoperation technology is being provided by a Portland, Oregon-based startup called Designated Driver. It will allow humans at Texas A&M to remotely control the shuttles in situations where the self-driving system may not be up to snuff, and they’ll also be able to interact with passengers on board. The new functionality could help solve a problem that similarly nascent autonomous shuttle programs have run into: crashes.
Multiple low-speed autonomous shuttles have gotten into fender benders
The low-speed autonomous shuttles currently whispering their way around a handful of downtown areas and campuses across the country are among the first real-world tests of self-driving technology. Since they’re extremely limited in scope — they typically ferry no more than a dozen passengers on short half-mile or one-mile loops — they serve more as an introduction to the idea of self-driving vehicles than as an exploration of how useful the technology could be.
Despite operating at low speeds on planned routes, a few of these shuttles have already gotten into fender benders. The first program to launch in the United States, in downtown Las Vegas, was clipped by a delivery truck on its first day of operation back in 2017. A similar service in New York City saw its self-driving shuttle bump into a car one day before the public launch.
So-called “safety drivers” or “safety operators” who ride on board are supposed to stop these crashes from happening, but in the case of the Las Vegas incident, the manual override controls were locked away. Having teleoperators standing by as a safety net is a potential near-term solution to this problem, according to Manuela Papadopol, Designated Driver’s CEO.
“Sometimes things don’t work. Systems fail. We’re still at the infancy when it comes to autonomy,” Papadopol says. She also says teleoperators may be in a better position to explain to passengers why a vehicle has stopped since they can see the full readout of the sensors that power the autonomy system.
Papadopol says nailing the basics of teleoperation now could solve other problems down the road, especially as companies look to remove safety operators from the vehicles. She says teleoperators could offer a “human” element by remotely instructing riders or serving as disembodied informational guides.
“Humans trust humans more than they trust machines.”
“Think about when you’re sitting in a plane on the tarmac, and you haven’t moved an inch in 20 minutes. There’s a moment of comfort when you hear the captain speaking to you, saying, ‘Hang on tight, we’re going to get there,’” she says. “Humans trust humans more than they trust machines.”
The associate professor in charge of Texas A&M’s autonomous shuttle program, Srikanth Saripalli, says he “realized there were lots of situations where we needed human input” during the first few months of the pilot, which ran for two hours a day on a half-mile loop in downtown Bryan. “So we started working with Designated Driver to include some remote supervision and, if needed, a teleoperation system in place, so we could compare how effective a safety driver is versus the teleoperation system.”
The beefed-up operation will take on a longer mile-and-a-half route, and the remote operation could allow for alternate routes, according to Designated Driver. Saripalli says the city of Bryan has access to the data from the program and will be able to make the call on whether Texas A&M can pull safety drivers from the vehicle.
Adding teleoperation to the mix means adding more technology to each of these vehicles, and it also involves increasing the number of humans involved in the near term — neither of which are good things if the idea is ever going to scale up while also staying cost-effective. Designated Driver’s goal of fully shifting the safety driver from inside the shuttle to a remote location could mitigate that, but it also opens up a host of new questions, including where will those jobs go? Who will be doing them?
For now, Designated Driver will run the remote operation out of Texas A&M, and it’s training members of Saripalli’s staff to monitor the shuttles. Determining how many shuttles one person can monitor at a time as well as what a full staff should look like at scale are parts of what the company aims to learn from the program, Designated Driver CTO Walter Sullivan says.
That said, the tech industry’s history with remote (and often contracted out) jobs is full of examples of the idea going sideways. If teleoperation takes off as a solution to some of the challenges facing self-driving cars, Designated Driver and the many other startups chasing this technology will have to answer those questions (and more) about who is doing the work and what that work looks like.