I'm reading all the comments criticizing FSD, and I honestly feel like we watch two different movies. My Tesla drives me from home to work, or from work to a downtown restaurant (granted, all within Silicon Valley where they probably have most of the data), mostly without my involvement. With the latest version, the car is occasionally too hesitant, and - very rarely - confused (e.g. by a railroad traffic light next to a road), but that doesn't subtract much from the experience.
I don't know of any other driving assist system that get anywhere close to this. Every time I tried fancy rental cars (BMWs, Audis, Mercedes) with the latest and greatest driver assist, it's feels like a joke.
There's a lot of confusion and people talking past each other about the difference between Autopilot, Enhanced Autopilot, the Full Self Driving option package, the feature called Full Self Driving beta, and the promised (but not delivered!) feature that will allow you to not pay attention, also called Full Self Driving. I think many people aren't aware of which is which and where and how they made promises about each.
It's almost like they intentionally picked these names to make them sound like far more than they actually are, and that's made it difficult for consumers to understand what they're getting...
Thanks - that might indeed be the case. What I'm referring to in my original comment is FSD Beta, which drives mostly by itself but requires constant driver attention (hands on the wheel, eyes on the road).
Self driving does not equal unsupervised driving (the phrase is not synonymous with autonomous driving). The same way that cruise control doesn't mean the car controls the cruising with no supervision.
What you’re missing are other people’s experiences. I completely believe your experience. I also agree FSD is fucking amazing compared to anything we’ve seen before.
However, driving in Houston, my Tesla would seemingly get in an accident nearly every single time I drive if I didn’t do something to correct. And at minimum it will do something stupid every single trip that causes a missed turn or exist that adds 5-10 minutes of driving. I would every single encounter it has around pedestrians on streets (no sidewalks in many parts of Houston) would have it wayyyy too close and going wayyyy to fast and make the pedestrian think I’m a huge careless asshole. There’s tons of other anecdotes, like it almost drive one of the wheels into a ditch (there’s tons of them in Houston). I would literally be hitting someone or something every single week if I blindly trusted FSD, and would an hour of pointless drive time every week too
This. The reason I stopped using FSD was due to traveling at speed on the biggest freeway running through Southern California when it randomly slammed on the breaks for no reason, until I quickly disengaged it. I’m not talking a gentle slow down. It went from 75 to 35 and all passengers getting thrown against their seatbelts and the car behind having to slam on their breaks hard to avoid rear ending me. First time, I thought it was a fluke. 2nd time, along the same route but a completely different place, the car behind had to swerve into the emergency lane to avoid running into me. I vowed never to use it again and join any class action lawsuit against the company regarding this. It’s dangerous AF on its best day.
Driver assist, however, is actually great and perfect for stop and go traffic. But damn that FSD is a literal killer.
Yeah, sounds like my Model S Plaid. Phenomenal performance, but autopilot is a joke.
I am dumbfounded every time I hear someone saying that their FSD is great, and that it doesn’t occasionally try to kill them or someone else. Frankly, I just don’t believe those stories.
Thank you for sharing your experience. It’s totally believable that FSD experience is very uneven across the country. Hopefully they will keep improving the system.
I think what’s missing is what the software allows. It could be BMW/Merc etc are way more conservative on what the allow the system to do and when they force the driver to take over. In certain contexts Merc is actually willing to assert and stand by a higher level of autonomy than any other manufacturer: (https://www.motortrend.com/news/mercedes-benz-drive-pilot-le...). Taking that at face value it’s possible they can do it and choose not to because they don’t want the liability. Whatever systems are in regular cars are then either borked or deliberately have less hardware.
Tesla is uniquely risk tolerant for better or worse. You also don’t hear about people getting into accidents in a BMW on self driving because they don’t make the same claims and have tons of safeguards.
> Mercedes says that Drive Pilot will only operate during daylight hours at speeds up to 40 mph on “suitable freeway sections and where there is high traffic density.”
> While the system is active, drivers must keep their faces visible to the vehicle’s in-car cameras at all times, but can turn their head to talk to a passenger or play a game on the vehicle’s infotainment screen. Drivers can’t crawl into the back seat to take a nap, for instance. The system will disengage if the driver’s face is obscured or an attempt is made to block access to the in-car cameras. Presumably the system will deactivate itself if it detects the driver is sleeping or operating the car while impaired.
<40 mph, specific freeways only, does not make any kind of lane change or exit autonomously. I think any carmaker with a decent off-the-shelf lane keeping feature could make a liability claim in this scenario. It's not a measure of the technology.
That you've convinced yourself of its safety based on your lone experience doesn't detract from all the other evidence that strongly suggests that the system is not safe.
Honest question, really - what do you actually like about it? I get that it's an amazing technological feat; it must be really wild to see your car just Knight Rider itself along. But... It's not like you can actually focus on other things, right? Is it really less stressful than just driving yourself there?
it is fun to watch the technology grow and morph- especially on the same daily route.
it’s also nice to have as an alternative to a knee when you need to do something awkward, like reach back or sneeze. there is always the heart attack scenario but overall i definitely trust it to drive for a short stint and to know it’s there to grab the wheel if i get in a bind.
on the highway it is excellent and really takes micro stresses out of a long drive. lane keeping is maybe 90% of it so i would weight this pretty low as a differentiator from other automaker offerings. still, i never had it on the car i came from- so it’s hard to sift that out of my impression.
Yeah, I have been driven to an airport in Boston without intervention. It’s like magic and only seems to be getting better based on videos posted every update.
If they'd have called it some marketing thing like "SuperCruise" or "HyperDrive" as opposed to "Full Self Driving" and "Autopilot," we probably wouldn't be having this conversation.
"It seems to me like Telsa's desire to settle was triggered by the court hearing date being set. Their initial settlement offer had three things I didn't like in it:
1) They only offered -L-5800 - but if I was successful in court I would get 8% interest on this amount from 1st Jan 2020 to the judgement, as well as court fees.
2) There was a "non-advice" clause, that I wouldn't help anyone with a similar claim to mine.
3) There was a confidentiality clause, that I wouldn't be able to discuss with anyone about the settlement."
He declined this offer and Tesla made a better one, more money and obviously no restriction on discussing the settlement and helping others.
Notably, Tesla countered by upping the offer but keeping the gag provisions. It was only when it was clear author wouldn't budge that they folded on those clauses. His conclusion was they really wanted to avoid the legal precedent (and attendant headlines). Tesla knows those crows will eventually come home to roost and the longer they can put it off the more they'll mitigate the blowback.
Here is a video here showing the Tesla Vision implementation of parking sensors [1].
After all these years if Tesla's ML models are incapable of basic static object detection then I don't see how they belong on cars at high-speed. And it says a lot about Tesla management that this made it into the car at all let alone as a replacement for USS.
I drove Teslas for 4 years and recently went back to a humble Toyota. I really don't think the pure-camera FSD will ever work properly. Between a Model 3 and Model X I've seen them do so many goofy things on the road, and get stumped by very common environmental factors such as fog, and strong sun with snow on the ground (glare). I've seen FSD get confused by poorly marked roads and swerve into exit lanes, or drive on the complete wrong side of the road. I stopped trusting it and I think cameras alone are just not fault tolerant enough.
I suspect the greater shortcoming isn't in the sensor suite department, but the intelligence department. Either way, what Tesla's putting on public roads currently seems criminally negligent and I simply don't understand why no TLA has intervened yet.
That decision was downstream of another mistake: selling a feature before the feature actually worked. They wanted to sell their cars as having all the hardware for FSD before FSD was actually implemented, and they wanted to make a tidy profit while doing it so they were strongly incentivized to conclude that the hardware needed for FSD would be cheap (excluding lidar.)
In 2016 Tesla claimed "as of today, all Tesla vehicles produced in our factory – including Model 3 – will have the hardware needed for full self-driving capability at a safety level substantially greater than that of a human driver".
It was false at the time and it's still false seven years later.
It was only a mistake if their goal is to be honest to their customers and provide the best possible product, but clearly other motives were/are involved ...
As of 5 or so years ago, the lidar used for this on the research side of things started at US$10k per unit (and for "surveying" you can do ok with one-on-top but for navigating you need something smaller and not sticking out of the roof, so then the car itself is a blind spot so you want 2x or 4x units.)
As a robot developer, I was really hoping that high-volume self-driving would lead to lidar mass production (and thus price drops), and there was one promising non-mechanical lidar that was aiming for $1000/unit for an self-driving-grade sensor, but I've lost track of it...
I wanna know how many of these refunds they've done...
I always suspected they'd refund FSD if you complained enough, and was surprised to not find a single person talking about a refund - so presumably the vast majority accept the gag order rather than take it to court!
I wondered if this was possible. Good job not caving on the gag clauses. Personally I'll keep FSD as it's actually useful sometimes and I like to test the improvements as they come out (responsibly). It's pretty much what I expected when I bought it (at a much lower price). But I could definitely understand if some people wanted refunds, especially if they bought at the recent high prices.
The difference is Tesla says you have to pay attention and enforces this, but it can attempt to drive on any street, make turns and lane changes, and navigate to your destination. Whereas Mercedes has a much, much simpler system that only works in a straight line going less than 40mph on certain freeways (so only in traffic).
Isn't this just for relatively low-speed driving, on freeways? IIRC it was limited to under 40 MPH, but on roads described in a way that it sounded like only freeways would qualify. I think there are lots of cars that can drive well under those conditions, though others won't assume the liability.
If you mean real self driving, I'd guess Waymo. I haven't ridden it, but the reports sound much better for it than the others. Possibly mobileye, but I've only seen company produced videos.
If you mean customer purchasable products, they all have various pros and cons. Out of spec motoring does a pretty good test called the "hogback challenge". GM SuperCruise tends to do well there, but there are lots of others that do well too.
I think your analysis is right. I also would say Tesla FSD beta is clearly the winner of customer purchasable products. It's worth pointing out that Waymo and Cruise very often fall back to remote operators when the car gets confused. This is no problem if you're evaluating how successful they'll be at operating a taxi service, but makes it more difficult to compare to consumer systems that do not work that way.
Readit News
I don't know of any other driving assist system that get anywhere close to this. Every time I tried fancy rental cars (BMWs, Audis, Mercedes) with the latest and greatest driver assist, it's feels like a joke.
What am I missing?
Here is how Tesla describes the differences: https://www.tesla.com/support/autopilot
Notably, they don't even talk about FSD beta there.
I think it's the other way around - the best software offered by conventional manufacturers (the best they can offer) is years behind Tesla's.
However, driving in Houston, my Tesla would seemingly get in an accident nearly every single time I drive if I didn’t do something to correct. And at minimum it will do something stupid every single trip that causes a missed turn or exist that adds 5-10 minutes of driving. I would every single encounter it has around pedestrians on streets (no sidewalks in many parts of Houston) would have it wayyyy too close and going wayyyy to fast and make the pedestrian think I’m a huge careless asshole. There’s tons of other anecdotes, like it almost drive one of the wheels into a ditch (there’s tons of them in Houston). I would literally be hitting someone or something every single week if I blindly trusted FSD, and would an hour of pointless drive time every week too
Driver assist, however, is actually great and perfect for stop and go traffic. But damn that FSD is a literal killer.
I am dumbfounded every time I hear someone saying that their FSD is great, and that it doesn’t occasionally try to kill them or someone else. Frankly, I just don’t believe those stories.
Tesla is uniquely risk tolerant for better or worse. You also don’t hear about people getting into accidents in a BMW on self driving because they don’t make the same claims and have tons of safeguards.
> Mercedes says that Drive Pilot will only operate during daylight hours at speeds up to 40 mph on “suitable freeway sections and where there is high traffic density.”
> While the system is active, drivers must keep their faces visible to the vehicle’s in-car cameras at all times, but can turn their head to talk to a passenger or play a game on the vehicle’s infotainment screen. Drivers can’t crawl into the back seat to take a nap, for instance. The system will disengage if the driver’s face is obscured or an attempt is made to block access to the in-car cameras. Presumably the system will deactivate itself if it detects the driver is sleeping or operating the car while impaired.
<40 mph, specific freeways only, does not make any kind of lane change or exit autonomously. I think any carmaker with a decent off-the-shelf lane keeping feature could make a liability claim in this scenario. It's not a measure of the technology.
That it could potentially kill you at any time.
That you've convinced yourself of its safety based on your lone experience doesn't detract from all the other evidence that strongly suggests that the system is not safe.
Last winter, when I was driving a lot on Spain highways, I really missed something like FSD or even a simpler Autopilot in my rental car.
https://youtu.be/7M-J3JydpPA
Tesla's "FSD" is nice in California I'm sure.
Do you feel like you are able to read HN while on the road or are you keeping “check” on the driving? If so, what have you won?
I like Tesla, but I really can't stand some of the things they overpromise, or some stupid choices like removing USS from their cars.
To be fair, I think that most of the things I am complaining about come from Mr. Musk choices...
1) They only offered -L-5800 - but if I was successful in court I would get 8% interest on this amount from 1st Jan 2020 to the judgement, as well as court fees.
2) There was a "non-advice" clause, that I wouldn't help anyone with a similar claim to mine.
3) There was a confidentiality clause, that I wouldn't be able to discuss with anyone about the settlement."
He declined this offer and Tesla made a better one, more money and obviously no restriction on discussing the settlement and helping others.
Some examples:
- When you drive in the night on a dark highway the car keeps complaining that the door pillar cameras are blind.
- Parking assist is a joke in calculating distances from other cars/obstacles. Also completely stops in poorly lit place. "Parking assist degraded"
- The emergency line crossed alarm sometimes triggers while I am dead center in my lane
- Blind spot monitor is basically unusable. Why would I want to look at the screen on my right when i try to change to the left lane.
- Windscreen wipers going on randomly when there is absolutely no rain, but don't start when there is light rain.
If they continue like this I can see BMW/Merc eating their lunch in a couple of years
After all these years if Tesla's ML models are incapable of basic static object detection then I don't see how they belong on cars at high-speed. And it says a lot about Tesla management that this made it into the car at all let alone as a replacement for USS.
[1] https://www.youtube.com/watch?v=kfsRFTpaTwA&t=1s
It's certainly not good they've picked systems strictly worse than human vision.
https://www.tesla.com/blog/all-tesla-cars-being-produced-now...
In 2016 Tesla claimed "as of today, all Tesla vehicles produced in our factory – including Model 3 – will have the hardware needed for full self-driving capability at a safety level substantially greater than that of a human driver".
It was false at the time and it's still false seven years later.
As a robot developer, I was really hoping that high-volume self-driving would lead to lidar mass production (and thus price drops), and there was one promising non-mechanical lidar that was aiming for $1000/unit for an self-driving-grade sensor, but I've lost track of it...
I always suspected they'd refund FSD if you complained enough, and was surprised to not find a single person talking about a refund - so presumably the vast majority accept the gag order rather than take it to court!
Slightly OT: What is currently the best self driving product the market has to offer? And what is it capable of?
If your Tesla crashes while using FSD, it's completely your fault, according to them.
If you mean customer purchasable products, they all have various pros and cons. Out of spec motoring does a pretty good test called the "hogback challenge". GM SuperCruise tends to do well there, but there are lots of others that do well too.
Deleted Comment