( I suppose going fully electric didn't preclude exclusive ev taxi use too )
https://medium.com/@trenteady/why-t...-in-self-driving-car-development-8eff18c65d3c
it would appear waymo disengagement 1 evey 40 miles ... so I couldn't yet get to work and back, and tesla apparently don't publish this data (weird),
especially data from the the busy school run with lots of targets to avoid, so it is unclear when machine driving singularity will be reached.
...but the software and maybe the Idar vs camera dichotomy is the key ? having a good range if you cannot watch netflix at the wheel, is not so useful
Because you’re comparing Apples to Oranges. Both companies are going about things totally differently tech wise, but they’re also at different points on the autonomous ladder.
Waymo are testing actual Level four autonomy, where disengagements are important. Tesla on the other hand have only rolled out (publicly at least) Level two tech, where disengagements will be much higher.
Tesla hope that their level two tech will “learn” faster than other companies, and as such eventually overtake the other companies.
The way they’re marketing their FSD and the way they have previously marketed their Autopilot is dangerous IMO, and that of others, including regulators who have chastised for it. Their FSD when announced as “feature complete” (presumably one the next few months) will be seen by many as Level four, but in reality it will still be level two for a long while yet.
Not only that, but Waymo and Tesla are at two different ends of the autonomy scale use wise too. So far Tesla have been working on the easy stuff, simple highways and recently slow, more complex areas (car parks). Waymo are in at the deep end working on complex situations in cities and towns. To put it another way, “highways” will have a dozen or so interactions on a trip, car parks will have similar, but towns will have hundreds, which is why most consumer ready products are for main road use only.
The argument Tesla are “ahead” is almost entirely based on the argument that their AI has more real world data to play with. True, but whether that data is actually more useful than simulations is up for debate, especially as other companies have shown a greater ability to function in more complex environments - worth pointing out here that the highway portion of their FSD has been around for about five years now and it’s still not up to Level three standard, let alone level four. The town portion they plan on releasing is going to take a lot longer than a year or so to get to level four...
