Waymo co-CEO Dmitri Dolgov has outlined why camera-only self-driving systems may not achieve full autonomy, citing "weak sensing" as a safety limitation. While cameras can approximate human driving or power assist features, Dolgov argues they fall short of the performance needed for true driverless operation. Waymo employs a combination of cameras, lidar, and radar, asserting that these different sensor types are not redundant but work synergistically to create a superior perception of the world. He highlighted scenarios where cameras are ineffective due to environmental conditions like dust storms or darkness, while lidar and radar provide clear data.

Dolgov also addressed the "nines problem" in reliability, explaining that achieving extremely high levels of safety requires exponentially more effort. He suggested that camera-only systems might plateau before reaching the necessary reliability for unsupervised driving. Regarding cost, Dolgov indicated that the price of sensors like lidar is decreasing with newer hardware generations, challenging the notion that they are prohibitively expensive.

This perspective contrasts with Tesla's camera-centric approach, "Tesla Vision." While Tesla's robotaxi service operates on cameras alone, Waymo's extensive testing with 220 million driverless miles reportedly shows a significantly lower crash rate compared to human drivers. Waymo is currently providing half a million paid trips weekly across 15 US cities and aims for a million weekly rides.