The car rolling down the window with no driver inside is just perfect.
I genuinely love it, not even in a snarky way. I sometimes get downbeat about it being freaking 2024 and we don’t have all the sci-fi awesomeness that I’d hoped for. But then there are moments like this. It’s the juxtaposition of the mundane and futuristic that sells it. Sci-fi authors hardly ever get that right. They’d put a self-driving car in, sure. But they’d never think of pulling one over because it drove into oncoming traffic, or that it would have an empty driver’s seat and wheel that turns itself, or that it would roll down the window so the officer could talk to tech support.
Here is a gift link to a New York Times article in which the writer compares the experience of being behind the wheel of a Tesla in self-driving mode to being in the passenger seat next to his fifteen-year-old son learning to drive.
I have a simple question but it may already have been covered but I am about a year behind in reading this thread. As most of you probably know there are self driving taxis in San Francisco now. Are these not considered “self-driving”? or have self-driving cars happened a lot sooner than many have predicted in this thread?
Apparently, there are people who are purists regarding “self-driving.”
If the car can’t drive anywhere on the planet by itself, they don’t considering it “self-driving.”
The fact that there are driverless taxis in specific cities doesn’t count.
If a driver in the car or a remote driver has to monitor the car and be able to take over at a moment’s notice, it’s not really self driving. It’s merely a capable driver’s aid. Someone still has to be paying attention and be ready to drive.
That is a pretty accurate assessment, and is the exact same analogy I (and possibly others) have used in this thread. Unlike many popular press on articles on self driving, that one seems pretty good. I’ve had my car self drive at the exact intersection he references.
He keeps saying that FSD (supervised) can disengage all of the sudden, which is true, but what is he doing that sudden disengagement causes a problem? He needs to be ready to take over at any time, without warning.
I’m also not sure what he’s doing that is making sudden, and unexpected, disengagement so common. In tens of thousands of miles I’ve had that happen probably a single digit number of times.
Do you have a cite for that? I have been trying to find information about remote drivers but all I find is mention of so-called “remote drivers” who don’t seem to have the ability to move the car remotely when there is an emergency and the fire dept. (i.e.) need the vehicle moved. Instead, this supposed “driver” send someone to the scene to manually move the car. To me that is not a remote driver.
I saw just this weekend someone jump out of a Waymo (I think) robo-taxi in crowded North Beach (SF). The car was in the middle of the intersection when the guy got out to argue with someone on the sidewalk. The car just stayed there and blocked traffic pissing everyone off. I’m pretty sure if there were a true remote driver they would have pulled it ahead and over. That didn’t happen.
Well, what they actually have is a place the car can call back to and get info from a human on how to proceed. It’s able to drive in a lot of situations, but it still needs input from a human to deal with situations it cannot. They call it “Fleet Response”
To me that is not at all a “remote driver”. It’s a dispatcher who sends someone out to move the vehicle.
from your link:
Much like phone-a-friend, when the Waymo vehicle encounters a particular situation on the road, the autonomous driver can reach out to a human fleet response agent for additional information to contextualize its environment.
That’s not a remote human driver actually driving the car. And that’s what I saw this weekend: A robo-taxi sitting there for 10 minutes blocking an intersection.
Well, if you’d read the cite, you’d know they don’t send anyone out (as far as the cite is concerned, I think they may have sent people out in the past). The remote assistant answers questions the robot has and can stop the car if that’s the safest course of action.
If a Waymo car gets too confused, it stops wherever it is. Regardless of whether that’s convenient or safe for other drivers. It “phones home” and a human who was paying no attention to that car will eventually have what amounts to a text chat with the car to tell it what to do. And just like you starting a text chat with “customer service” at your bank’s website, there may be lots more customers / car wanting to chat right now than there are human agents available to clear the backlog. So you / the car waits your / its turn.
Once the car gets useful advice, it drives off if it can. If not, eventually a human is dispatched to drive it or tow it. So overall, not a real-time right-this-minute solution.
Big picture, this is not that different from a human driver having a stroke. Except the Waymo brakes to a halt as it recognizes it’s stroking out, while the human probably unknowingly stomps on the gas and hits a storefront or crowd of peds at high speed. A car stopped in an intersection with a dead or incapacitated human driver will be there for a long time before emergency services has been notified, arrives, removes the person, and tows the car. The last case like this I was involved with was about 45 minutes from start to finish. Then traffic could start moving normally again.
All this is darned inconvenient for other drivers. The question I don’t know is how often per mile do Waymos go stupid versus humans going stupid from stroke, heart attack, drunk, falling asleep, too busy texting to look outside, etc.
Fairly accurate and matches more or less what all of us regular FSD drivers have been saying. I don’t think it’s quite that bad and it’s much better than that in certain situations. You certainly wouldn’t text or stop paying attention when your kid is driving but it’s really nice when they can take you to an event an hour away through freeway traffic. Again, you need to learn to use FSD. It’s not that big a deal.
Looks like 12.4.2 wasn’t fully baked either, but 12.4.3 may be the one. About 12% of AP4 cars and 4.5% of AP3 cars have it now: https://teslafi.com/firmware
Sadly, that doesn’t include me. Anyone else get the update yet? Word (from what I gather) is that it’s somewhat more cautious than 12.3.6, but it is an improvement in intervention rate, and the handsfree driving is nice. Looking forward to it.
I ran across a blinking red signal the other day. Wasn’t sure if the car would (correctly) treat it as a 4-way stop, but it did fine. On the other hand, there’s a particular intersection that keeps confusing it. In a left turn lane, there’s a KEEP CLEAR section ahead of the crosswalk, because pulling too far up intersects with a light rail track. Nevertheless, the car pulls right up to the sidewalk. It actually does so in a safe way by hugging the outer edge, but it’s still a problem because it’s past the sensor for the signal. Not a problem if there’s other traffic, but there wasn’t.
I made the decision to jump to the UI track so I’m on 2024.20.3 because I eventually want 2024.26 which has native YouTube Music. I’ll be on FSD 12.3.6 until the next merge. I don’t mind the steering wheel nag that much.
The wheel nag is not that bad. I’ve found the bare minimum needed to keep it happy–just a little wiggle every few minutes. And I can spot the indicator out of the corner of my eye. But I still want the update .
.