Is this very impressive? SF to LA, unless you're traveling on local roads, is essentially all freeway. And pretty much one long road at that. Hop on I-5 or 101 or even the 1 and, in the words of The Simpsons, "sit back and feel your ass grow" for five hours. Self driving folks have been saying that highways driving has always been pretty straightforward for essentially a decade now. Perhaps I'm being cynical, but "completing an extremely well-traveled, uncomplicated route without error" doesn't seem like much of an accomplishment in the year 2021.
This video is interesting because it covers both dense urban streets and freeways, shows unedited footage including the autopilot debugging view, and is done with someone's personal car (as opposed to a company test car).
These driver assistance is IMO especially useful onn such boring drives. I don't expect this tech to be legally cleared for busy city streets for a long time.
The #1 use case of self driving for me is long distance automation. It should also be the easiest for freeways.
Besides automating my current long highway drives, it will make the decision for any trip under 300 miles an automatic drive vs fly.
For 400-600 miles, 90% chance of driving if I'm paying or staying longer than a carry-on.
And I cannot wait for self driving ev RVs that will drive placed while I sleep in back, even if I am strapped in. I might sell my house and go full Chris farley
The article title is misleading. There was an intervention in LA to avoid debris on the road (“suddenly revealed” after car moved out of the way) and another case where the car acted incorrectly but the driver chose not to intervene so they could have an intervention-free trip.
(This is according to the article body; I didn’t study the video.)
I imagine that spending hours letting the car drive itself while only making sure the car doesn't do something stupid might be less enjoyable than driving for hours.
Not only do I agree that it would probably be less enjoyable, but I also expect it would be less safe.
A driver actively piloting a car will be focused and experienced at that task. That same driver will not have had any tuition or assessment on their ability to judge whether a car with advanced assistive piloting* is acting sensibly.
In the case of some event requiring manual intervention, I'd be surprised if the driver's focus would be significantly worse than a similar event while actively driving. Not only that but the judgement, knowledge, and experience of being able to identify such events in a timely manner.
I'm genuinely shocked that this feature is allowed on public roads. It's one thing a car owner consenting to being a trial participant, but pedestrians and other road users have no such opportunity to provide or deny consent and are at equal or greater risk if things go awry.
*I refuse to use Tesla's irresponsible 'self-driving' marketing nonsense
