A comparison of Waymo's and Tesla's autonomous driving architectures across sensing, representation, prediction, planning, validation, and training. Waymo relies on lidar, radar, and cameras feeding a Foundation Model that produces structured, inspectable representations, verified through separate Teacher/Student models and a Safety Framework, reporting 220.6 million rider-only miles and published crash-reduction statistics. Tesla relies on a pure-vision camera approach with neural networks trained on fleet telemetry from millions of consumer vehicles, verified currently by human drivers under Full Self-Driving (Supervised), with its own Vehicle Safety Report methodology. The piece frames the core tradeoff as how much knowledge is fixed in advance (Waymo) versus computed at drive time (Tesla).
Table of contents
Matic: The World’s First Intuitive Home Robot has arrived. (Sponsored)SensingRepresentationPredictionPlanningValidationTrainingConclusionQuestions this post answers
How many rider-only miles has Waymo driven and what safety improvement does it claim over human drivers?
Waymo reports 220.6 million rider-only miles through March 2026, meaning no human occupied the driver's seat, across five metro areas. Measured against human crash rates in the same operating areas, adjusted for where within each city the service drives, Waymo reports a 94% reduction in serious injury or worse crashes and an 82% reduction in injury-causing crashes. Track how autonomous vehicle safety claims are measured and compared by following this space on daily.dev.
Does Tesla's Full Self-Driving use lidar or radar, or only cameras?
Tesla's Model 3 and Model Y run camera-based Tesla Vision without radar or lidar, relying entirely on a pure-vision approach using exterior cameras and neural network processing. This contrasts with Waymo's sixth-generation system, which carries 13 cameras, four lidar units, six radar units, and audio receivers for redundancy in poor conditions like rain or ice. Compare sensor-based architecture decisions like this one on daily.dev when evaluating autonomous systems design.
How does Tesla measure the safety benefit of Full Self-Driving (Supervised) compared to manual driving?
Tesla's Vehicle Safety Report compares Teslas with Full Self-Driving (Supervised) engaged against manually driven Teslas using the same telemetry pipeline, reporting 7 times fewer major and minor collisions and 5 times fewer off-highway collisions. A collision counts as system-engaged if it was active within five seconds beforehand, and Tesla attributes no fault in the reported data. Follow developer-focused breakdowns of safety benchmarking methodologies like this on daily.dev.