-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Autopilot Questions, Experiences, Problems, Concerns
- Thread starter yobigd20
- Start date
I have tremendous interest in autopilot. I also have tremendous interest in abs brakes and stability control. All will save lives.
Everything suggests that for some number of years you will absolutely be able to retain complete control of the car if you want, with the possible exception of some emergency conditions such as an impending collision with the car in front of you. After automatic systems are proven to be much safer than human, you might have to get a special license to drive manually. The tests for this license will become ever more stringent, and eventually the liability issue will be turned on its head, with manual driving and human reflexes considered too dangerous to the drivers around you.
In the meantime Tesla is going to be collecting reams of data from you car on how well their algorithms are performing with respect to your actual driving. No other manufacturer is going to have that much sensor data.
Last edited:
http://seekingalpha.com/article/2748755-tesla-claims-are-out-of-sync-with-reality
from the article (fyi I hate seekingalpha articles, but this one raises points that I've also had concerns about):
the third bullet there makes me think the author is a Tesla hater therefore his opinion is biased but whatever I guess there is some truth to that statement that their communication sucks.
so really it's about the lane change. if you go back to my OP my first bullet point "there's potential for the system to accidently cut someone off and they'll smash into you."
I think I've discussed this further before, but can't find it in this thread. It may have been another thread. but anywho, the point that I made before was that there is no long range radar in the rear. there's only one and it's forward facing. there is no way for the system to detect a very high speed car coming up in the lane next to you. I've already heard the arguments, that it's up to you to always be in control anyway. I think my points still valid though, that for a good lane change system there really needs to be long range radar on both corners of the rear bumper (or higher up somewhere, highest point on the sides maybe, something like that).
from the article (fyi I hate seekingalpha articles, but this one raises points that I've also had concerns about):
the third bullet there makes me think the author is a Tesla hater therefore his opinion is biased but whatever I guess there is some truth to that statement that their communication sucks.
so really it's about the lane change. if you go back to my OP my first bullet point "there's potential for the system to accidently cut someone off and they'll smash into you."
I think I've discussed this further before, but can't find it in this thread. It may have been another thread. but anywho, the point that I made before was that there is no long range radar in the rear. there's only one and it's forward facing. there is no way for the system to detect a very high speed car coming up in the lane next to you. I've already heard the arguments, that it's up to you to always be in control anyway. I think my points still valid though, that for a good lane change system there really needs to be long range radar on both corners of the rear bumper (or higher up somewhere, highest point on the sides maybe, something like that).
AutoPilot != autonomous.
I expect AutoPilot lane change to be:
(1) turn on turn signal
(2) press and hold turn signal in same direction
The idea is simply that combined with good LKA lane changes become simple, smooth and precise.
It's not going to detect traffic coming from the rear, but in the future that car coming up from behind will automatically brake.
green1
Active Member
if Tesla was advertising a system that changed lanes without input then I'd agree with you. But they require the driver to manually initiate a lane change. That means it's the driver's responsibility to do it right. Unless or until Tesla sets the vehicle up to do it without input there is nothing wrong with them not having those sensors.
This is fine for auto-pilot, but not for self-driving. but of course Tesla has never claimed self-driving, so it's a non-issue.
This is fine for auto-pilot, but not for self-driving. but of course Tesla has never claimed self-driving, so it's a non-issue.
DDotJ
Member
I drive back from work on the 55 Southbound freeway towards Newport Beach and when I reach McArthur, every time the lane departure warning goes berserk and the steering wheel vibrates like crazy because it thinks that I am leaving lanes whenever the pavement changes color (newly paved black asphalt on the left and gray colored on the right side). It also drops the speed limit to 45 mph despite the fact that i'm in a 65 mph zone. I'm afraid what kind of havoc this will bring when AutoPilot is introduced... hopefully there are some software enhancements to improve lane guidance and speed limit sign warning. maybe GPS assisted?
I would agree that you need some form of moderate range detection to the rear. I don't think it needs to be radar, and it probably doesn't actually need to be as long ranged as the detection to the front - in any lane change type situation the overtake should be substantially less than your current speed, though it's best to leave margin.
The current backup camera may be a possible solution - it's pretty wide angle, but also very high resolution, so it may be practical to do object detection and closure rates off of it.
More likely, though, is that for safe fully autonomous lane changes you need something more - like maybe the rear view cameras that Tesla is planning to put on the Model X (and I suspect may find their way back to the Model S at some point.) Those should be more than adequate to the task - they're nominally the tool a human driver would use to decide if there was a problem, and the computer can certainly analyze the data much faster.
There are a number of production adaptive cruise control systems that run off of the data from a single visual camera, with no radar (like the one in the BMW i3,) so the code is certainly out there and fully developed.
Walter
With all this talk of Tesla liability for autopilot mishaps, (from a forum of enthusiasts no less) my biggest fear is that some idiot will fall asleep, or be involved in a heads down romp through America, and have the unattended autopilot drive through a mall at high speed. The vultures will be looking for very deep pockets, and Tesla will suffer in court at the hands of a low information jury. Tesla lawyers will scream to the engineers, and the next OS pushed to our cars will be sans autopilot.... forever.
I've spent decades using the most sophisticated autopilots ever developed, yet if mine steers my aircraft into the abyss because I wasn't paying attention, and somehow I live to tell about it, the Captain is at fault, not Boeing. It is our responsibility as owners of this incredible car to use the autopilot as a tool to aid the driver, not as a crutch, and not as a babysitter. And, if the worst happens, and you find yourself wrapped around a tree while the autopilot was "steering", tell the vultures NO when they crawl out of the woodwork en' mass to represent you in their dream lawsuit against Tesla. Do this for your fellow enthusiast.
I've spent decades using the most sophisticated autopilots ever developed, yet if mine steers my aircraft into the abyss because I wasn't paying attention, and somehow I live to tell about it, the Captain is at fault, not Boeing. It is our responsibility as owners of this incredible car to use the autopilot as a tool to aid the driver, not as a crutch, and not as a babysitter. And, if the worst happens, and you find yourself wrapped around a tree while the autopilot was "steering", tell the vultures NO when they crawl out of the woodwork en' mass to represent you in their dream lawsuit against Tesla. Do this for your fellow enthusiast.
Canuck
Well-Known Member
Disagree. Fault can be apportioned and if the autopilot on any plane or car steers someone into the abyss then the makers of the autopilot are at fault. That does not mean the driver or pilot is not also at fault for failing to correct it, but that alone does not absolve the manufacturer of liability.
There are many countries where you can't sue or hold individuals or corporations liable for fault, and I am always glad, and feel much safer, when I return home from those countries.
I wonder how this case turned out. Enquiring minds want to know!
- - - Updated - - -
Not true,,,, The pilot is always responsible for the flight of his craft. Autopilot is just a tool to aid in the pilots workload. The autopilot is a mechanical device that will fail. The pilot has to monitor what this tool is doing at all times.
- - - Updated - - -
Not true,,,, The pilot is always responsible for the flight of his craft. Autopilot is just a tool to aid in the pilots workload. The autopilot is a mechanical device that will fail. The pilot has to monitor what this tool is doing at all times.
Canuck
Well-Known Member
No - "the pilot is NOT always responsible for the flight of his craft". If the engines fail, the pilot is not liable.
However, I never said the pilot is not liable if autopilot fails. I agree the pilot may also be liable. But fault is not mutually exclusive. Once again, fault can be apportioned. I agree that "Autopilot is just a tool to aid in the pilots workload." If, in fact, "The autopilot is a mechanical device that will fail" then the reason for that failure may be as a result of fault in the way the device, it's hardware or software, was manufactured, installed, programmed etc. If it's determined that a reasonably prudent company would not have made the errors that the caused the failure, then fault can be found with the manufacturer in addition to the pilot. It happens all the time in our courts with product liability claims.
Here's the problem with that statement:
"...the logic of the automation is fundamentally different from the logic the pilot would use to control the aircraft. Thus, once the automation is engaged, the pilot cannot meaningfully participate in the decision making process. Rather, he can only monitor the results that the process generates. That makes it difficult for the pilot to detect errors until after the aircraft has deviated from the desired course. Other studies note that, once the pilot has detected an error, he frequently becomes pre-occupied with “fixing” the automation, rather than controlling the airplane. Perhaps the fixation is due to the pilot’s unwillingness to admit that he doesn’t fully understand the automation or, more accurately, the automation’s logic. Perhaps as a “monitor” rather than a participant, he is uncomfortable with “jumping back in” and hand flying the aircraft because he has, to some extent, lost situational awareness.
Some studies suggest the solution is to provide pilots with better training. At least one author notes, however, that to understand and thereby promptly remedy the many errors that that the automation can produce, the pilot needs to understand the system logic to a degree that is simply not practical."
General Aviation Automation Risk
Last edited:
The pilot can be liable if the engines fail. Again, a mechanical device that will fail at some time. Depending on the pilots actions he could be a hero, or at fault, if it is determined that he did not take appropriate actions when they failed
http://en.wikipedia.org/wiki/US_Airways_Flight_1549
Captain Sully was hailed as a Hero, however there were some actions that were not normal. He took control over the actions of the copilot who was in control. He did not close the vents prior to a water landing that assisted in the plane from sinking following the water landing. So even though he was labeled a hero, he still made mistakes! The outcome made him a hero...
Last edited:
Canuck
Well-Known Member
Yes, I agree. I was more making the point that the pilot is not always responsible for the flight of his aircraft. It all comes down the specific facts of each case.
By the way, good to speak to you again Lloyd. I hope all goes well.
By the way, good to speak to you again Lloyd. I hope all goes well.
You are reporting actual experiences with Autopilot on the Model S, not some other car? If so, that is great, and I hope to see other actual reports in this thread. I understand that this subject leaves a lot of room for speculation because much of the implementing software will roll out over time, but I appreciate actual reports. Has anyone posted pictures of the new front-looking camera? Do we know where the radar is? What is its range ?
A friend of mine just took delivery of a Model S that has the autopilot hardware. I will see him and his car this Saturday. Does anyone have suggestions as to what to look at and/or test?
I know someone who has some other brand of car with lane assist features. She reported it mistakes shadows of power lines for lane markers, as well as the pavement color changes that you mention.
yesterday I got the red banner across the top of the display, front camera blocked. Seems to happen in rain. I find lane departure works as good as my BMW one. The speed assist is good but should be percentage above the speed limit. +10 on a 35mph limit is probably a ticket. +10 on a 70mph limit is maybe not. Overall I like it and I am looking forward to more features coming.
DDotJ
Member
Yes, this is from my Model S which is about a month old. This is NOT Autopilot, just the beginnings of the Autopilot features, aka lane departure and speed assist. The radar is the little black square under the nose cone, and the front facing camera is the triangle shaped cutout where the rearview mirror mount meets the windshield.
Very experienced Major Airline a Captain here.
The Captain is always responsible for whatever happens to the aircraft, but it may not be his fault that it happened.
The Autopilot is a tool to reduce fatigue and workload, the crew need to monitor and adjust what it does to achieve what the goal is. I see the Tesla autopilot the same way. When I cruise on a freeway I may turn it on but I will monitor the operation. If I change lanes, for example, I will check for traffic before I change.
The Captain is always responsible for whatever happens to the aircraft, but it may not be his fault that it happened.
The Autopilot is a tool to reduce fatigue and workload, the crew need to monitor and adjust what it does to achieve what the goal is. I see the Tesla autopilot the same way. When I cruise on a freeway I may turn it on but I will monitor the operation. If I change lanes, for example, I will check for traffic before I change.
Similar threads
