Look from a consumer standpoint why would I want a system like this? I mean aside from the toy factor, why would I trust it? I can’t take a nap while it’s driving. I can’t look at a screen or read a book. I have to in fact be more on alert than when I am driving because when I am driving I know what my inputs into the car are and I just need to worry about the external inputs (shape of the road, traffic, signs, road conditions). With this system I also need to guess what the car’s brain will do which is a lot harder. This would be like always driving with a nervous teenager who is actually operating the car but expects you to take over at any moment because they might mess up. And for the price difference would at some point it be cheaper to just hire a driver?
I would love something like this for the longer, boring, predictable drives.
For example the I5 drive from LA to SF. That's a pretty straightforward, predictable and mostly uneventful 6h drive that seems straightforward to delegate to a machine. There are no pedestrians, no intersections, nothing crossing the street. The main dangers would be around objects on the road, other vehicles getting in a crash or doing something unpredictable next to you, animals running onto the path of the vehicle. As a human you might actually be worse equipped to react to in time compared to a car.
That one use-case alone where the car drives itself on the I5 while I read a book or watch a movie would be worth something. The human can handle the quirks of the more treacherous local roads. I'd still prefer a bullet train of some kind, but in its absence, this would be a decent climate-friendly alternative that doesn't require a giant plane getting up in the sky for a 50 min flight.
I've done enough really boring driving to find that about once in every N hours (depending on the road) a very NOT boring event will occur, requiring immediate and correct response. I don't think any road exists without that happening.
Yes this is making the assumption that, when the unexpected occurs, an experienced human driver still outperforms whatever self driving code may be running.
Not even Tesla is saying their system could or should be used that way. You're making quite a few assumptions about the capabilities of the system and going outside of the instructed and permitted use of the system if you are going to go read for 6 hours while the car drives itself.
Unexpected situations can and will happen. They are not at all as simple as "well the car will just detect it and break faster than me" by a long shot. What if there is debris on the road and the car misreads it? What if there are actually pedestrians because somebody got into an accident and decided it would be a bright idea to take a little walk?
I could go on, and I'm sure you or somebody else could answer every hypothetical I could come up with. That's not the point. The point is that even a "boring, predictable drive" is anything but. We put in a ton of processing power that makes that drive (usually) seem boring and predictable. A level 2 system is not designed to be let loose even in these circumstances.
You'd be required to pay attention continuously and be able to do immediate takeovers at all times for that LA-SF drive at L2. You simply wouldn't have to actually drive. L3 is where your attention is "allowed" to wander. Quoting NHTSA guidance:
> Therefore, the L2 driver is expected to be alert and monitoring the road continuously. The L3 driver is not expected to be paying attention to the road at all times, but is expected to be takeover-ready with advance notice.
The included autopilot seems good enough for this usecase, many report it makes long drives less tiring.
Level 2 FSD honestly seems worse from a human UX standpoint that it seems like an anti feature to me. I'm a fan of tesla and I've come around on autopilot being a nice feature but I think this FSD focus is silly and overhyped at least in regards to providing any value to customers anytime soon.
That would be a level 3 system (ie one where the driver can disengage but be ready to react immediately when the car tells them to). This is a level 2 system; ie you must be alert at all times in case the car tries to kill you. You can’t watch a movie in a car like this.
Slightly misleading headline. The Tesla spokesperson said the 'Auto Steering on City Streets', which is the formal name of the feature currently in beta, will remain hands on, basically because it's illegal to do anything otherwise. That doesn't mean they aren't pursuing FSD and won't release it to the public, but that it needs to be done with regulatory approval.
I hate how the passenger is absolutely amazed and always commenting how well the car handles some minor tasks while it fails miserably at double parking, turning, and most importantly driving predictably for other cars.
The driver is luckily a lot more skeptical but that could be attributed to the fact the he is the one being highly alert and keeping the car from running into things.
I think the video works well with the two of them there - it lends some legitimacy to a video that could otherwise be dismissed as he's obviously enamored with the idea while the driver is more skeptical.
I initially read it as “FSD will always be Level 2” but they’re just talking about the beta:
> "[A] final release of City Streets will continue to be an SAE Level 2, advanced driver-assistance feature," continued Williams. "Please note that Tesla’s development of true autonomous features (SAE Levels 3+) will follow our iterative process (development, validation, early release, etc.) and any such features will not be released to the general public until we have fully validated them and received any required regulatory permits or approvals."
As I read it, it’s even more specific - they’re talking about the ‘City Streets’ package - and suggesting that the functionality with this particular name will stay <=level 2.
This as they develop further functionality, maybe >level 2x they’ll name it something different.
(It’s silly how even anti-Tesla articles fall foul of the ‘anything Tesla-related is worthy of an article’ click-bait approach. There really isn’t anything to see here.)
Much as I'd love to have L5 self-driving today, it would probably be at least a little funny to see the number of people on this board insisting it would be available in 2018 or 2019 and certainly not many years out.
It probably is funny, although it is still infuriating as fuck to read people claiming that there is no need to invest in public transportation because self driving cars are right around the corner.
Well, and if we could wave a wand and have universal self-driving--while remote work throws something of a wrench into the calculus--you presumably have people living further away from work, having a car circle the block a few times rather than park, etc.
I'd place a bigger and bigger bet on cars driving fully autonomously on designated sections of highway and Manhattan being essentially a maybe someday location over the next decade or two.
I think this title and article are overstating things, are clickbait:
>"[A] final release of City Streets will continue to be an SAE Level 2, advanced driver-assistance feature," continued Williams. "Please note that Tesla’s development of true autonomous features (SAE Levels 3+) will follow our iterative process (development, validation, early release, etc.) and any such features will not be released to the general public until we have fully validated them and received any required regulatory permits or approvals."
City Streets is a feature of the current offering, not what Tesla apparently intends to deliver in finality.
It's very strange these days to see headlines about a self-driving setback and be happy about it. I always wanted it as a kid but ever since someone pointed out to me that self-driving cars is the end of personal car ownership I hope it never gets here. Why would a company sell me a car when it can use that same car to generate ten times the income as an autonomous taxi?
