Key facts
- Waymo argues that safe, fully autonomous driving at scale requires a combination of cameras, lidar, and radar.
- Waymo asserts that Tesla's camera-only, end-to-end AI approach risks 'black box failures' and 'hallucinations'.
- Tesla is preparing to launch its Cybercab robotaxi, which lacks a steering wheel or pedals.
- Waymo has expanded its robotaxi network to over a dozen U.S. cities.
- Tesla plans to produce over 125,000 Cybercabs annually.
- Waymo's multi-sensor approach is more expensive than Tesla's AI-focused strategy.
Waymo has publicly challenged Tesla's approach to autonomous driving, asserting that a combination of sensors, including lidar and radar, is essential for safety at scale. This critique comes as Tesla prepares to launch its Cybercab, a purpose-built robotaxi that relies solely on cameras and AI. Waymo's VP of driving software, Srikanth Thirumalai, stated in a company blog post that "safe, fully autonomous operations at scale require more" than just cameras, citing over 200 million real-world miles of data. He also warned that Tesla's "pure end-to-end" neural architecture approach risks "black box failures" and "hallucinations."
Tesla CEO Elon Musk has previously dismissed lidar as a "crutch," advocating for a camera and AI-centric system. The Cybercab, designed without a steering wheel or pedals, is slated for a September 3 launch, with Tesla aiming for an annual production of over 125,000 units. Despite years of development and recent trials in Texas and Florida, Tesla still needs to demonstrate its software's capability for full autonomy at scale.
Waymo, which operates a robotaxi fleet in over a dozen U.S. cities, utilizes a more conservative technological approach integrating cameras, lidar, and radar. This method, while more expensive, has allowed Waymo to scale its operations. The differing technological philosophies set the stage for a significant competition in the burgeoning robotaxi market, with Waymo questioning the viability and safety of Tesla's strategy.