Waymo, like Uber and others, are not really building self driving cars. They are building something much more like self driving trains. Trains not bound to a route by steel rails, but by a high precision map. If they go where no high precision map exists they are lost. If the world changes too much from the map they are lost.
Tesla, like other current luxury car makers, offer something that can roam freely using a few basic heuristics about following lines and avoiding things that show up visually or on radar. These systems have very different strengths and weaknesses from the Waymo system.
Before being too impressed by Waymo's claims I would need to see a LOT more information than they have published to date.
Waymo do seem to rely heavily on detailed map. I don't fully understand why but it's been suggested it's simply easier and therefore safer. Not sure about all the other teams though.
It's not a big problem for Waymo because Google are already mapping all roads for
Google Street View regularly, they just have to add more sensors to those cars and then they will have detailed maps of virtually everything.
(And if you think about it, we also rely on internal maps kept in memory (to some extent) when we drive in a familiar area, so maybe it's not so strange after all.)
Before being too impressed by Waymo's claims I would need to see a LOT more information than they have published to date.
Iv'e been following their progress since the days of Stanley and I guess I have gained some trust in some of the people involved. I'm feeling quite confident they will surpass expectations. They don't seem to cut corners or rush things but keep playing it safe, and they are doing everything the same way I would have (based on my limited knowledge, but still). (Or maybe I'm just a bit of a fanboy?)