Chris Urmson apparently just said (or its being claimed he said) his stuff from 2010 was more advanced and or further along than Tesla's current FSD beta tech and that he didn't think the Tesla stuff would work for a very long time if at all. Really(?) well Urmson was in 2021 back in 2011 so he's had an extra decade to get it right. Where is it then? I think the little gig that floats these companies is claiming they made some deal with some ICE company to inflate the value of ICE stocks. They shell game some money and that is all it amounts to. Why? In part because ICE companies don't want self-driving because it means they sell less units because less units are needed with people opting not to own cars anymore because robo taxis are cheaper- ICE companies with their sunk costs don't survive that. But they are also threatened by Tesla so they have to put out some bs purporting they have options (when they don't) and its always in the form of some press release saying some company gave Cruz money or its GM's rag Navigant saying GM is the best or its bought and paid for CR saying it. But none of it has any reality it all just appears to be lies. Thing with Waymo and Google... it would have been here already but they certainly realized their geo-fenced lidar approach wasn't going to be economical against what is coming. Their secret sauce which they never really let the public in on seemed to be using remote humans to supervise around the edge cases the way the military uses humans to supervise drone kill decisions from across the globe. So they took the human drivers out of their vehicles but were just remote driving their semi-autonomous drone vehicles as necessary the way the military has done for a long time. That doesn't really scale when true autonomy is on your heals. And it is really a less taxi drivers model or partial not full automation. They still had their talk of number of driver interventions but they why did they have to keep the supervision part? They still had a good record but their system wasn't really logical or efficient dropping people off on the wrong side of the street and avoiding certain types of turns to pad things-they hit a wall. Thing about a pure camera approach. There are nuances that don't get brought up much that might give a better feel. It is not just for instance that humans solve the driving problem visually and cameras give all the critical contextual information. The cameras in the cars for the purpose of driving are better than human eyes in a lot of ways that go beyond having 360 vision. My sense is they do better in the rain and (that goes beyond the phobic coatings and shapes that keep clean) and better in bright sun directed or glaring light and better in total darkness (modern cameras don't need star light amplification or any fancy military stuff like that- they can just plain see in the dark even in color and have done that for a long time- cheap Cannon hand held's since mid 2000s did this) and will do better in the fog and AI is amazing at inferring and predicting from clues we have no idea about just on vision and of course it constantly vigilant without gap. But I think at this point if they use Lidar they aren't serious and can be dismissed- it means they are still at the level of not understanding the problem after Tesla has solved it. They have lost and it will soon reflect in their valuation and that of the ICE makes, they won't be able to ponzi shell game their way out of the correction more propaganda- it catches them and wipes out and transfer their value and makes people resistant to giving them access to capital again. The comment from Urmson seems totally panicked desperate and bitter. There may be a Chinese company with a solution somewhere but I strongly sense there is no ICE maker with access to a solution.
