On a sunny June afternoon in Chandler, Arizona, more than a dozen police and emergency vehicles paced up and down a mostly empty street, with their sirens blaring and emergency lights flashing. All eyes, though, were on the handful of self-driving cars that shared the road. Some drove in front of a fire truck; others cruised alongside police motorcycles and unmarked cars. Spectators, including Police Lieutenant David Ramer, watched in anticipation as the cars decided when to pull over and when to yield.
For the hours-long demonstration, Ramer spent three months coordinating with Alphabet’s Waymo to find the right time, place, and fleet of vehicles to help the company train its cars to recognize and respond properly to emergency vehicles. This is a task most human drivers have yet to get right, and still, it’s an essential one for machines to master before they can take over on the road.
“Emergency vehicles [in Arizona] are supposed to stay in the fast lane on the left, and everyone is supposed to slowly pull to the right,” Ramer says. “Very, very rarely does that happen; most people panic and park right there.”
So he welcomes the department’s partnership with Waymo, which according to a 43-page safety report published this month, is creating a “library of sights and sounds” to feed into its software.
As more autonomous and semi-autonomous cars find their way onto public roads, law enforcement officers and first responders are figuring out how to handle them in collisions or during traffic stops. It was only two years ago that a police officer in Mountain View, California, pulled over a car only to find that there was no driver inside (there was a passenger, however). The incident invited jokes and snickers on the internet, but also raised serious questions.
“How do those first responders know if this thing is an automated vehicle rather than a driver vehicle, and how do you know that it is fully off?” asks James Hedlund, a consultant for the Governor's Highway Safety Association who previously worked at the National Highway Traffic Safety Administration. “How do you know what it's doing, what it’s thinking?”
Since then, there have been several, more serious incidents that have called police, firefighters, and paramedics to the scene, from a high-speed collision in Arizona that flipped an autonomous Uber vehicle on its side to a fatal crash in Florida between a truck and a semi-autonomous Tesla.
In fact, Hedlund says, the hardest questions come even before the arrival of fully autonomous, so-called Level 5 vehicles. (Most have only reached Level 2, meaning the car can automatically steer, accelerate, and decelerate, but a driver has a to be alert and ready to take control of the wheel. Recently Audi introduced the world’s first Level 3 AV.)
“It’s Level 3 that is the real tricky one,” he says. Much like in a Level 2 vehicle, the system in a Level 3 AV will steer and accelerate on its own. But in this case, the driver won't need to monitor the road under certain conditions, allowing him or her to be truly hands-free. “Was the driver informed that he or she had to take over?” he continues. “If so, did the driver really take over?
At least one company is actively working with first responders on those questions. In the same report that highlighted Waymo’s partnership with Chandler, the company also offers a glimpse at its collaboration efforts with fire and police departments in cities where they test their cars. Those cities, according to the report, can get on-site training for their officers and first responders on how to identify and assess self-driving cars after a collision, as well as “a line of communication for further engagement.” (Waymo has not responded to CityLab’s request to elaborate.)
Click here to read the full article: https://www.citylab.com/transportation/2017/10/waymo-police-first-responders-emergency-vehicles-driverless-cars/543351/