How self-driving Ubers respond to Pittsburgh’s awful weather

Sure, Pittsburgh’s hills and bridges were appealing for testing. But we also have four seasons.

Jasmine Goldband / The Incline
MJ Slaby

Stand close enough to one of Uber’s self-driving cars, and listen for a puffing sound.

That’s the cleaning system, said Brian Zajac, head of hardware engineering for the company’s Advanced Technologies Group, while answering questions during Cafe Sci at the Carnegie Science Center last week. We wanted to know more: how exactly do autonomous cars handle Pittsburgh’s awful weather?

To figure out what weather condition to tackle first, Zajac said Uber studied the most common weather scenarios that Uber riders in the human-driven ride shares encounter. That’s rain, not surprisingly. It can be unpleasant to catch an Uber in a downpour, Zajac said.

The cleaning system uses puffs of air to keep the cameras and sensors on the self-driving vehicles clean, and that’s especially helpful for rain droplets, he said.

Leaders of the Uber ATG have repeatedly said Pittsburgh’s challenging hills and streets made it ideal for developing self-driving cars that can then easily navigate other cities. That goes for the city’s weather, too, Zajac said. Having four seasons is helpful for working on day-to-day operations and mapping, he said. Over the winter, those working on self-driving cars wondered about the impact of snow, but the winter ended up to be a mild one.

Zajac said expanding the car’s capability to drive in snow, ice and fog, as well as to clean bug splatter from its cameras are all things Uber is working on. Heavy hail and dense fog will be an interesting technical challenge for engineers, but they’re lower on the list than more common conditions like nighttime operations in dimly-lit areas or places without street lights, he said. Issues that impact the most people come first, and that influences how Uber thinks about temperature.

Automakers are held to high standards when it comes to withstanding environmental conditions, Zajac said. Those are “great standards and benchmarks,” he said, adding that Uber first wants to develop self-driving cars that navigate areas where the most people are, not to withstand extreme heat or cold.

As the cars advance, Zajac said Uber staff will take the roles of air traffic controllers or central dispatch at a trucking company. The conditions that planes can’t fly in are well known, he said, adding that it will be the same for self-driving cars, where Uber staff are paying close attention and adjusting the number of cars or the time they spend on the road based on conditions.