On average, self driving car technology is faced with a lot of challenges such as more than 5 million crashes take place annually and 22 percent of them are weather-related, the Federal Highway Administration (FHWA) reports.
Now with more autonomous vehicles being deployed on highways around the country, researchers are working on ways to fine-tune and enhance the technology that controls these vehicles. Capabilities the cars are being taught include identifying cyclists, maintaining speed limits and learning traffic patterns. Tackling the challenge that bad weather poses is just one step of the process.
With mass-market availability of electric cars around the corner. Self driving car technology such as in the Waymo’s cars will be faced with teaching these cars how to operate reliably in scenarios that don’t happen often in real life and are therefore difficult to gather data on. Anything from strange weather occurrences and other vehicles’ unpredictable driving patterns to more common but irregular situations, like emergency vehicles and snowfall, can pose a data problem. Without consistent opportunities to encounter these situations during average training sessions, self driving car technology often must undergo specialized training for those scenarios, which takes time.
The vast majority of weather-related crashes occur on wet pavement and during rainfall, the FHWA states. In California for example, opportunities to drive in wet weather have been limited due to the state’s five-year drought.
“Driving in rain makes many human drivers nervous due to reduced visibility, and some of our sensors, particularly the cameras and lasers, have to deal with similar issues,” Google stated about their electric car. “Our laser sensors are able to detect rain, so we have to teach our cars to see through the raindrops and clouds of exhaust on cold mornings and continue to properly detect objects.”
Google also revealed that its cars have built up a “library of various sirens” and taught its software to identify them, so when an emergency vehicle approaches, the cars will “drive more conservatively until it has a better sense of where the sirens are coming from. The autonomous cars are also equipped with cameras that can detect flashing lights so if an emergency vehicle is coming through an intersection, the car can stop and will resume when it is safe.
Waymo has also began training its autonomous Chrysler minivans to be able to recognize what they look and sound like in real life situations. Alphabet’s autonomous vehicle spinoff has teamed up with Chandler Police and Fire in Arizona to set up an “emergency vehicle testing day.” The authorities had ambulances, police cars, motorcycles and firetrucks pass by, trail and lead the Chryslers all day and night while the minivans’ sensors collected as much data as possible from all speeds, distances and angles.
To train vehicles for rarer scenarios, the data collection gets even harder. In the course of reporting on the impact of automation on trucking towns, Quartz’s Mike Murphy and Dave Gershgorn spoke with a trucker who explained how the broad sides of a truck made it prone to tipping over without experienced maneuvering in the wind patterns characteristic of the western US. Given the erratic nature of the wind, it is also difficult to model the self car driving technology. “There are too many variables,” said Terry, a fellow trucker.