What do you call the person who babysits a driverless vehicle? The driver? The passenger? He’s called a safety conductor–not a bad title for Robert Haas, who in his youth wanted to be a train conductor.
Haas is a conductor for one of the first two driverless vehicles to offer rides to the public in Ann Arbor. Operating on a route around the U-M’s North Campus Research Complex since June, they carry members of the U-M community from NCRC Building 10, near Plymouth Rd., to U-M parking lot NC91, on Baxter near Hubbard. The “Navya Autonom Shuttle” is deployed by Mcity, the university’s autonomous vehicle test facility.
Haas’s wife, Anne, spotted the job posting last year and knew it would appeal to a guy who loved trains and has taught driver’s ed. Carrie Morton, deputy director of Mcity, says they needed “people who are detail-oriented, can focus, and can adapt to technology.” Haas was a perfect fit.
Hired in November, he trained with five other conductors at Mcity’s sixteen-acre site, which includes roads with intersections, traffic signals, sidewalks, simulated buildings, streetlights, and obstacles such as construction barriers. Haas says driving there was “not quite as harrowing as cruising in my own car in downtown Ann Arbor, but is a fair simulation.”
Now, at 8:30 every weekday morning, he walks to a parking bay at NCRC to start up his “automated and connected” all-electric shuttle for its one-mile tour around the complex. “I use a portable device sort of like an Xbox to input the route, I check the charge to be sure it’s 100 percent, and check if all seven computer systems are working.”
He carries a horseshoe-shaped control device that’s designed to be gripped with both hands. His thumbs operate two joysticks, for steering, accelerating, and decelerating, and buttons that make the vehicle go, stop, steer, ring the bell, open and close doors, and operate lights and turn signals–“all except doing the dishes,” Haas jokes. “But most of the time, the vehicle does all this automatically. My job is to monitor error messages generated by the shuttle, watch outside for objects or people in the way, and be aware of unexpected slowdowns.”
He describes the vehicle, manufactured by the French firm Navya Technologies, as “a big oblong bubble, with windows all around.” There’s no front or back–each end has the same functions and appearance, allowing it to go forward or backward without turning around.
Haas wipes a bit of dirt off the side, radios dispatch to confirm his departure, then hops in. The eleven seats are arranged in a circle, facing one another like a mobile support group.
Ideally, if the shuttle were operating at full implementation, there’d be no conductor. But for now, Haas uses the control console to manually steer out of the parking bay onto Baxter.
—
Seventy percent of drivers are uncomfortable sharing the road with driverless vehicles, according to a survey reported by CBS. Mcity researchers want to know how passengers, pedestrians, and other vehicles react to driverless shuttles to gauge consumer acceptance of the technology. Though the shuttles are operated by U-M Logistics, Transportation & Parking, at this stage they’re officially a research project, not a primary transportation service.
The shuttle detects and avoids obstacles with lidar, which uses invisible laser beams to build a view of the surrounding environment and paints the footprint for its route. It scans 180 degrees side-to-side and forward as far as two football fields. With its own 3-D picture of the environment, it can tell which objects require a stop.
The route map is depicted on a touch screen mounted behind Haas’s head. He selects Route 1–to the NC91 parking lot–and Route 2, back to Building 10. He presses “go,” which rings a bell, and off it goes.
At 9 a.m. Haas picks up his first load of passengers. U-M students, employees, faculty, and their adult guests can just hop on. Many are what Haas calls tourists–engineers and technical people who want to experience the shuttle. Anyone not affiliated with U-M needs permission from Mcity.
“For their first time,” Haas says, “it’s like Cedar Point–they’re apprehensive but excited. They want to know who is steering. I like the passengers. They ask good questions. I get to meet people from around the world.”
The shuttle detects bikes and pedestrians, slows for a person at the edge of a crosswalk, and stops for a person in its path. “So far no one has emerged ashen faced!” Haas says. “The most frequent comment: ‘It’s a little slow.’
“The only person who was unhappy was a man who brought his child, and I had to tell him that the published rules say no kids are allowed. I had to use a whole lot of charm and my considerably modest diplomatic skills.”
Haas hands out cards asking riders to take an online survey. In addition, “After every shift I send in a report on unusual events and performance issues, like if it slows at a certain place, as well as traffic issues. I also report on how people interact and are excited about their ride or comments like, ‘I can walk faster than this.’
“Every day there are things you wouldn’t anticipate. This is a real test. But there are other challenges.
“Weather, for example: Snow is not an issue if it’s fine and small. But the lasers love reflecting off of white things like big snowflakes. They confuse the laser systems and cause the vehicle to slow repeatedly and stop every few yards. Protocol is to cease operation if it snows.”
Driving resumes once the snow stops. The traction is good–the battery-powered shuttle weighs more than 5,000 pounds and has four-wheel steering. The positioning system is so precise that on runs after a snowfall, the shuttle’s tracks exactly match the ones it made on the previous run.
“There are many more variables for the shuttle to confront,” Haas says, “lots of pedestrians, traffic all day, parked cars, delivery trucks like FedEx, and U-M maintenance vehicles. And the pizza guy. Unexpected real-life events are all over you.”
When it encounters an obstacle the shuttle stops six to eight feet away and blows its horn every fifteen seconds–and will keep doing so, says Haas, until “the cows come home … Once we had an inexplicable slowdown from a manhole cover.”
When that happens, Haas can assess the problem, then start up again. And he can see some things the shuttle can’t. “One day a baby groundhog was crossing. The laser can’t see below six inches, so the passengers watched while I manually paused the shuttle to a smooth stop to let it pass.
“The shuttle goes only twelve miles an hour, and often vehicles behind me are not happy, since the posted limit is twenty-five. At stop signs the university asks that I stop and start manually. Since the vehicle takes eight to ten seconds to resume and proceeds slowly, people pass illegally all day, so we have to stop. I understand that the researchers are trying to make the stop-and-go speeds shorter.
“Sometimes when we are making a legal left turn and have the right of way, the car behind thinks we are too slow so tries to pass while we are turning left. While the shuttle would detect this and stop itself, I’ve been trained to err on the side of caution, so I push the emergency stop button on the console even before the shuttle responds.
“I make the passengers wear seat belts because when I push the red button on the console the brakes lock, and it stops on a dime. As we approach a crosswalk, it slows and stops if a person steps out. You could say it obeys the local ordinance.
“Three times pedestrians who like to live dangerously have jumped in front, and the shuttle braked itself with its emergency stop. They jumped back in time then took off and laughed. We got tested and passed. That’s a victory for me.”
