Waymo's Dmitri Dolgov: 20 Million Rides and the Road to Full Autonomy

Autonomous vehicle development requires a transition from rapid experimentation to a rigorous, safety-first engineering culture. Waymo’s approach centers on the "Waymo Foundation Model," an end-to-end architecture that integrates a driver, simulator, and critic to achieve superhuman performance. By augmenting learned representations with structured, materialized data, the system ensures high-fidelity validation, currently operating 13 times more safely than human drivers in serious injury-causing scenarios. This technical foundation supports a shift from early-stage R&D to rapid, parallel global commercialization across multiple cities. Beyond the software stack, the sixth-generation hardware suite emphasizes performance, cost reduction, and rider-centric design. Ultimately, the mission remains focused on reducing the 1.19 million annual global road fatalities by proving that autonomous systems can reliably navigate complex environments, such as detecting pedestrians obscured by buses through sparse Lidar sensor returns.