For Waymo’s robotaxi service, future expansion depends on a critical challenge: mastering winter driving. Having initially focused on warmer climates like Phoenix and Los Angeles, the company is now targeting East Coast cities such as Boston, New York City, and Washington, D.C., where snow and ice are unavoidable.
At a recent company meeting, Waymo’s chief winter weather expert stressed that for the company to enter new markets, its autonomous vehicles must handle snow confidently and safely.
Robert Chen, the company’s product lead for weather, highlighted the urgency of this goal. “This winter season is gonna be a really important season for us,” he stated. To compete with human-powered ride-hailing services that operate year-round, Waymo cannot remain a fair-weather-only service. “We really aspired to build this product and the service that people can rely on… not just eight or nine or ten months out of the year, but something that they can really rely on and use whenever they want,” Chen explained.
Like human drivers, autonomous vehicles perform best in clear conditions with dry roads. However, snow presents unique difficulties. According to Phil Koopman, an autonomous vehicle expert at Carnegie Mellon University, snow can obscure road markings and signs. While a human can infer a stop sign from a partially visible shape, “machine learning can have trouble with that if it hasn’t been trained on partially obstructed stop signs,” Koopman said.
He noted that Waymo’s multi-sensor system, which includes lidar, radar, and cameras, is better equipped for these challenges than camera-only systems. “Cameras are gonna have a lot of trouble with blowing snow,” he explained, adding that “radar’s gonna really help.”
Beyond the technical hurdles, Waymo faces a significant data challenge. Chen noted that snowy conditions are rare in the company’s existing dataset, sometimes accounting for less than a fraction of a percent of its total driving data. This scarcity has prompted the company to employ advanced AI and simulation to augment its training. Waymo has also been accumulating real-world mileage in snowy climates, including Truckee, California; Michigan; and Upstate New York, with further testing in Denver and Seattle.
While Waymo’s current fifth-generation system can handle cold weather and light snow, its upcoming sixth-generation Driver is being specifically engineered for severe winter conditions. Hardware improvements include miniature mechanical wipers to clear the rooftop lidar sensor and more powerful heaters to defrost all sensors. The system is also trained to handle icy roads and share data across the fleet. If one vehicle encounters a slippery patch, Chen said, “It’ll actually send that information to the rest of the fleet and now other other vehicles in the fleet know that that particular location is slippery.”
In cases where conditions become too hazardous for most drivers, Waymo may temporarily pause its service. When real-world testing isn’t possible, the company relies on advanced simulation to replicate rare weather events. Waymo is integrating generative AI and layered models that can distinguish between different types of snow—such as wet, powdery, or slushy—to continuously refine its training pipelines.
Although customers may have to wait for their first snowy robotaxi ride, the company is making progress. Waymo plans to launch in Washington, D.C., next year, with future expansions planned for other East Coast cities, London, and Japan. As his team prepares for the coming season, Chen acknowledged the scale of the task. “The self-driving problem is really hard on its own,” he said. “Now you add in these crazy weather conditions. It’s a pretty challenging task.”
Source link
