Waymo’s Data-Driven Journey in Building Autonomous Trucking

Autonomous trucking is a high-stakes industry. Truck crashes cost $30 billion annually, the U.S. has a shortage of 80,000 truck drivers, and approximately one-third of trucks drive so-called empty miles, or the mileage that accrues while driving an empty container or trailer and not earning revenue. Building a more capable driver is essential to safer roads and is one of today's most complex technical challenges.

Waymo Via's head of trucking, Boris Sofman, shares his insights about how the Waymo Via team is building a better autonomous truck driver, which includes maintaining a single-driver platform across passenger vehicles and commercial trucks. Sofman also discusses how his company leverages synthetic data to improve the driver in rare scenarios, including different weather situations, construction scenes, and sensor degradation. He also covers his team’s use of advanced sensor suites, data collection, and labeling techniques, including using multiple vehicles for better ground truth.

Prior to Waymo, Boris was the Co-Founder and CEO of Anki, where he and his team developed various AI-based entertainment products and shipped over 3.5 million robots and devices around the world. In other positions, he also worked on off-road autonomous vehicles and ways to leverage machine learning approaches to improve navigational capabilities in real time.