You are buying the hype. Waymo is only good enough to play the odds of some good tech. That will run out.

The fact is no one in this industry is using a viable development approach. No one can spend the lives, money or time to get close. Each of these need to be resolved to reach L4:

-Non-existent General Learning - This results in little inference. That in turn results in an untenable scenario, object etc pattern recognition effort from a safety, time and cost POV. Also, "pixel" or micro level classification is often wrong when only a few pixels are in contention. Like shadows on a stop sign. Or a shirt pattern.

- Use of human Guinea pigs, many of which have to literally be sacrificed to train/test many crash cases. Additionally, it is virtually impossible to regain situational awareness in time critical scenarios, many of which are crashes.

- Relying on the real-world vs simulation for most development and testing

- Use of inadequate gaming simulation technology vs that from aerospace. And no Waymo does not have the right sim tech or modelling approach, especially regarding sensors and the perception system.

- Sensor design needs to include a camera, LiDAR, imaging radar, sound location and possibly thermal. LiDAR needs to help by creating tracks and aiding classification. Imaging radar needs to help with classification and localization. It all needs to tie to a Kalman filter that auto-adapts to the environment