-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Don't take your hands off the wheel
- Thread starter smatthew
- Start date
Well, again, to each his own but if it was only tunnel, that would be great but what came up recently in the media about the the system not seing the semi makes me wonder even more.
Of course, I realise that one must not take his eyes of outside but then, it should not be called an auto pilot.
Knightshade
Well-Known Member
It certainly is.
Just sadly most people are not pilots and don't understand how limited many actual autopilot systems have been especially early in their development.
I see your point, however back in the time the first APs were being built and used, the technology was not at today stage.
Also if the car AP is not approved for some situations, it shouldn't engage and it should disengage when the situation is encountered. For instance, in a lot of planes which are not approved for automatic Take off, you can try engaging the Auto pilot bellow 400 feet but I can guarantee you it won't do anything.
It is. I'm a pilot too, although I fly private planes, small single- or twin-engine personal aircraft with 4-8 seats. Essentially all autopilot technology in airplanes is intended to reduce your workload by taking care of repetitive or "boring" chores like holding altitude or course, thus letting you focus on being the pilot instead of the driver... thinking about weather, systems, wind, passenger comfort, fuel, and of course monitoring/supervising the autopilot!
The only real difference is that an airplane's autopilot has a far easier job: no potholes, and the "lane" is 8 miles wide and 2,000 feet high. No sharp turns, no irregular markings, no weird surface problems, and no need to avoid traffic all the time that is < 1 sec from potentially causing an accident. Tesla's autopilot naming may not be intuitive to most, and there's clearly a marketing issue there, but it's actually a great name from a purely technical point of view: it does things for you and lowers your workload, but is NOT self-driving.
All of the above should not be taken to mean I'm satisfied with current progress. The car's environment is far more challenging, and Tesla's autopilot is still clearly deserving of its "beta" status. It makes mistakes which it shouldn't, and it's not as good as it should be, but then that's why it's in BETA. Whether a beta product should be released to the public is a different debate in which I don't care to participate. All I'm saying is yes, it's an autopilot; and yes, considering the challenges it faces and its BETA status it's pretty good, though not yet even close to good enough for non-beta trust-it-with-your-life release. We're years away from that.
Fernand
Active Member
I don't understand the conflict here. If you drive a Tesla car
with EAP or with the current FSD upgrade, you learn what
it's capable of, and where not to rely on it. It's unimaginable
that anyone who drives such a car would be going back to
the dictionary or some company literature to decide what
the terms mean so they can adjust their behavior. It's all
part of a continuum of automation, from the first horseless
carriage, to automatic transmission, cruise control, and now
the gradually improving iterations of self-driving technology.
The term Autopilot cannot have a precise meaning in the
face of changing technologies. It started with a rope on a
boat's rudder, and keeps evolving, with growing pains at
every step. What's the issue? Rope slipped off the rudder?
Personally I agree with Elon Musk's expectations. I have
no doubt, zero, that in a few years we will look back at
manually operated vehicles as senselessly dangerous
machines from the days of the uniformed hotel elevator
operator and scythe-cut wheat crops.
I'd say fans of the good old days or those who focus on
the growing pains should just use only what they prefer
and stay out bashing arguments. It may be sad, but at
some point, for safety reasons, they will be restricted to
the horse and bicycle lanes.
Also, as more vehicles are more automated, they will be
more predictable, thus lowering the risks. Maybe as an
interim measure all cars will carry a cheap transponder
to help other vehicles more easily "see" them.
@antoinearnau do you drive a Tesla with NOA capability?
People who would prefer an automation system that self
disengages when it encounters unfavorable conditions
have no idea of what they are asking for. Not only is it
unnecessarily restrictive, it's also far more dangerous to
pull driving assistance out from under the driver. I know.
with EAP or with the current FSD upgrade, you learn what
it's capable of, and where not to rely on it. It's unimaginable
that anyone who drives such a car would be going back to
the dictionary or some company literature to decide what
the terms mean so they can adjust their behavior. It's all
part of a continuum of automation, from the first horseless
carriage, to automatic transmission, cruise control, and now
the gradually improving iterations of self-driving technology.
The term Autopilot cannot have a precise meaning in the
face of changing technologies. It started with a rope on a
boat's rudder, and keeps evolving, with growing pains at
every step. What's the issue? Rope slipped off the rudder?
Personally I agree with Elon Musk's expectations. I have
no doubt, zero, that in a few years we will look back at
manually operated vehicles as senselessly dangerous
machines from the days of the uniformed hotel elevator
operator and scythe-cut wheat crops.
I'd say fans of the good old days or those who focus on
the growing pains should just use only what they prefer
and stay out bashing arguments. It may be sad, but at
some point, for safety reasons, they will be restricted to
the horse and bicycle lanes.
Also, as more vehicles are more automated, they will be
more predictable, thus lowering the risks. Maybe as an
interim measure all cars will carry a cheap transponder
to help other vehicles more easily "see" them.
@antoinearnau do you drive a Tesla with NOA capability?
People who would prefer an automation system that self
disengages when it encounters unfavorable conditions
have no idea of what they are asking for. Not only is it
unnecessarily restrictive, it's also far more dangerous to
pull driving assistance out from under the driver. I know.
Last edited:
That is exactely how it works with planes, and it removes all doubt on what is going on.
Of course it gives a warning when it disconnect so you know that there is something going on, which in return raises your awareness.
I tested it during the trial.
Fernand
Active Member
Ah, I see. So you are not using NOA now? But then you're fine!
And this whole debate's about whether you should have gotten it,
which you didn't ?
And this whole debate's about whether you should have gotten it,
which you didn't ?
Wizard2416
Member
I agree to disagree.
My previous vehicle before Tesla M3 was a 2017 E300 with distronic plus (which is the driver assistance package) and have to say that it is way inferior than the EAP package that Tesla offers. The Mercedes Benz package was not able to take over the turns and in some cases it disengaged while driving which was not that pleasant specially when doing turns.
Have to remember that Tesla's AP is not FSD. Therefore it has limitation and it's learning from its own mistakes. With that in mind, any claim about AP not performing as expected is probably expected due that is a Beta version plus its stated in the manual and the alert message you get when engaging it.
I do love my Tesla and don't regret switching but I do also understand it's limitation and that if we want a better product, we need to train it.
Yeah absolutely! FSD appears to be diluting down to what previous EAP was aspiring to be. Certainly nothing like a literal full self-driving car where you can sit back and pay no attention to the road. Achieving level 5 automation is not just a "software" engineering task, it involves many other engineering disciplines in parallel e.g. vehicle dynamics. Tesla have already historically way underestimated the timeline for FSD, so I have little confidence in their current marketing talk.
I love my Tesla MX, we have an M3 on order as a second car and think current EAP is really cool, but I have realistic expectations for its capability. If I didn't keep my hands on the wheel in EAP it would have taken me out at least half a dozen times over the last 18 months of ownership! It is improving for sure, but still a mountain to climb.
That's the dumbest diatribe I've heard in awhile. You've clear!y never worked in software.
Please provide more detail of your aviation experience to understand what you are speaking of. I would say that light, general-aviation airplanes certified to FAR 23 and under 6000 or 7000 pounds are the only appropriate comparison to a car, and even 7000 pounds is a fairly large, pressurized twin. And on those airplanes, unless they are equipped with a radar altimeter (which most are not), the autopilot does not have a clue how far above the ground you are and this "will not engage below 400 feet" idea does not apply.
In general, a performance comparison between aircraft autopilots and a Tesla AP software is not fair... the airplane's job is much easier. But even then, it sounds like maybe you're speaking of the autopilot technology in larger, more sophisticated aircraft, and that's an even less applicable comparison as you're then talking about aircraft that, even used and 80's vintage, are likely to be $2M and up.
I am flying a 737, but I don't think money is the matter. If the car AP is not aporoved for city driving, it shouldn't engage in city. The car has a gps hence it knows where it is.
strider
Active Member
Then why did Elon promise a coast-coast FSD drive this year? Well, he's promised one for the last few years....
Oh good grief. Look at my .sig line. I have been involved with Tesla WAY longer than you. I've never owned nor shorted a single share of Tesla stock. I am simply a long term customer calling it like it is. As long as people treat the company with kid gloves they will have no reason to improve. Further, if random people talking on an internet forum can tank a stock, then the stock is overvalued.