Self-driving cars are still learning unwritten road rules, like the Pittsburgh Left
The Pittsburgh Left might be not in the rulebooks, but you better program any self-driving cars driving in the Pennsylvania city with that intel, because it's going to happen.
Uber is one of several autonomous vehicle companies testing in Pittsburgh that's had to contend with the unwritten rules of the road that can sometimes manifest in aggressive ways. Drivers' tendency in Pittsburgh to turn left before oncoming traffic has a chance to cross the intersection gave this dangerous move its moniker. As a result, Uber now has to train its cars to expect the abrupt left turn.
This Tweet is currently unavailable. It might be loading or has been removed.
While Uber, Aurora, Waymo, Aptiv, and other autonomous vehicle companies need to keep up with legit city- and state-specific laws and build software systems that can follow the rules no matter where the car is driving, there's still a whole set of cultural and societal norms to follow. If a self-driving car doesn't drive like the locals, it stands out even more —and not in a good way.
You May Also Like
Amazon-backed Aurora programs several social behaviors that aren't the law, but makes cars with its self-driving system get along better with others on the road. Some examples include turning right on red in states where it's allowed. This gets tricky because cars aren't legally required to turn right on red, but tell that to the cars piling up behind you as you wait for the green.
When merging from two lanes to one, Aurora is training its system to practice the "zipper method" of every other car. Again, there's nothing legally stating cars have to do this, but common road courtesy prevails. When on the highway, Aurora is keeping its vehicles from sitting in the left-most lane unless it needs to pass. Highway patrol won't ticket you for bad passing practices in some states, but you'll annoy your fellow drivers anywhere, so the software is getting trained on how to pass courteously.
And yes, like Uber, the Aurora autonomous vehicles are learning how to handle unprotected left-hand turns, especially in situations that come up in Pittsburgh and other cities.
This Tweet is currently unavailable. It might be loading or has been removed.
Self-driving startup Aptiv has cars in Las Vegas, Boston, Pittsburgh, and Singapore —and soon China. It's powering the self-driving Lyft rides available in Vegas. As Aptiv's president of autonomous mobility Karl Iagnemma noted, each place dictates different driving behavior.
"We have to build a system that allows us to adhere to the rules of the road of every city," he said. "And different driving norms and preferences."
It could be something as simple as Singapore drives on the left-hand side of the road or Vegas has dust storms the car has to deal with. In Boston, because they test by the water, a car once approached a flock of seagulls on the road. While unordinary for human drivers, the cars need to know that because it's in a sea-side environment birds may appear and the perceived obstacle will fly away. But in other location that mass in the road could very well be something to avoid.
"No one wants to develop a technology that only works in one city," he said.
A recent Aptiv project looked at yielding in different cities and how long cars take to wait before heading through traffic. At junctions they measured different yield times and found a sweet spot where back-seat passengers felt most comfortable. For some places when the car moved through at a quicker pace passengers reported feeling like a human was driving, instead of a reserved, overly cautious machine.
It's about bringing that human touch to a machine-driven vehicle. "It's hard to get a car to drive in a human-like manner," he explained, while also staying safe and law-abiding. After looking at the data, Aptiv can program its vehicles to be a typical driver for a city, but it's important to add skill, safety, and a more confident vehicle on the road.
Cruise, the autonomous vehicle company from General Motors, is testing its electric Chevy Bolts on the streets of San Francisco. While hills aren't so much a challenge for the self-driving system, the unique cable car trolleys and tracks on the city's peaks bring in unwritten rules.
Cable car riders tend to hop on and off whenever and certain streets give cable cars priority. Cruise has to be able to detect that it's driving near a cable car and anticipate erratic human behavior. Simply following the rules won't cut it.
Topics Uber
Sasha is a news writer at Mashable's San Francisco office. She's an SF native who went to UC Davis and later received her master's from the UC Berkeley Graduate School of Journalism. She's been reporting out of her hometown over the years at Bay City News (news wire), SFGate (the San Francisco Chronicle website), and even made it out of California to write for the Chicago Tribune. She's been described as a bookworm and a gym rat.
