Jul 22, 2015
AmpedRealtor My car has no such features as are being discussed here. Since my purchase, I have learned how Tesla releases software - with many bugs - and I am extrapolating that experience to Autopilot. The only safety critical features on my car are airbags and ABS brakes. I would have no reason to believe those systems don't function properly - they've been around for decades and the logic for those systems is probably not even coded by Tesla. But Autopilot is unexplored territory by a company under the gun to deliver the software, and programmed by people who are working far too many hours. You can't account for every variable in actual driving conditions, and guaranteed there is at least one scenario that Tesla has not thought of, or coded for, that could kill someone. Again, this is just my opinion.
You don't know you have a problem until you have a problem. This happened with the battery puncture fires, and ultimately Tesla had to re-evaluate the ride height as well as the lack of battery protection. I don't want to be one of those guinea pigs.�
Jul 22, 2015
brucet999 One wonders how they could have done that without the next generation of sensors on the car?�
Jul 22, 2015
Ugliest1 IMHO, they were able to do that using smoke and mirrors. Or CGI.
I *cannot believe* they are promoting email-checking, movie-watching, and complete driver inattention.�
Jul 22, 2015
mfreese From the Youtube video caption:This is a promo video produced by Bosch, not Tesla. I would venture to guess it is largely CGI.�
Jul 22, 2015
Johann Koeber
They didn't.
The Bosch test vehicle is full of sensors and electronics still in a developement phase.�
Jul 22, 2015
PatD It's pretty easy to see those screens are simulated. They just did it to show what's possible in the future.�
Jul 22, 2015
brucet999 Yes, I later noticed a second video showing stop-action of 1500 man-hours of work to install many meters of wiring and numerous sensors being installed.�
Jul 22, 2015
Ingineer I did notice they made quite a few changes to the interior of the car.
Here's the video with the work shown:
Note lots of new sensors, including stereo cameras!�
Jul 22, 2015
Footbag Not really similar. The nav in the IP looks slightly like v7... dark and 'round', nothing else looks remotely similar (aside from the dark color scheme). While I didn't go through/see many of the options of the touchscreen, what I did see was very similar in overall function/look as what we have now, whereas the video is quite a departure. I could have certainly missed something, and elements could change before release.
- - - Updated - - -
You may not be a happy camper then :-( But a agree, much of the Bosch concept is intriguing.
Oh, and hello 691HP, it was nice meeting you today at the Burlington Supercharger!�
Jul 22, 2015
Ingineer Here's the "real" test car:
�
Jul 23, 2015
brianman Thanks for sharing the video, Ingineer.�
Jul 23, 2015
mgboyes This is a CGI-enhanced promo video for an R&D division within Bosch which is working on fully autonomous self-driving. Once that arrives (10 years from now; maybe sooner) then obviously you'll be checking your email, watching TV, having a sleep, etc while the car drives for you.
But it has nothing to do with what Tesla are planning for a look & feel overhaul in their v7.0 software update.�
Jul 23, 2015
R�B This part bugs me. You're building a $100k machine, give it a $300 desktop class processor and be done with it for the next 5+ years. Why such a underpowered Nvidia Tegra SoC solution that was used to power mobile tables 2 years ago. The response times of the UI are my BIGGEST pet peeve when I drive my car everyday. Either the software is insanely un-optimized or the hardware is underpowered. With such a large amount of the "quality" perception of the vehicle coming from the User Interface, Tesla should never have shipped software this slow....�
Jul 23, 2015
Ingineer This bothers me too. They built the center display modular with a sub board containing the CPU, ostensibly so it could easily be upgraded and at relatively low cost. The Tegra 3 was arguably obsolete even before the first MS was delivered, especially when you consider the size of the screen. This is no different than the cellular modem, which is also on a sub-board. The just recently started shipping cars with LTE modems instead of 3G, so it proves they are willing to do some upgrades. Why not CPU? I suspect they intentionally don't want the browser too fast or capable, because we'd be using more bandwidth on their dime. It's also arguable a safety feature; The browser is so slow that you don't really have to take your eyes of the road that often to read it's updates! =)�
Jul 23, 2015
artsci Agree. I have no interest whatsoever in autopilot, none.�
Jul 23, 2015
Stoneymonster I don't even own a TV! :wink:�
Jul 24, 2015
Gizmotoy It's the "hardware is underpowered" option. Nvidia was in trouble with the Tegra and likely offered Tesla a sweetheart deal to get someone, anyone, to buy it. Is easy to forget now, but the Model S's future was legitimately uncertain when they were choosing these parts. Plus, they were trying to get under $50k at the time, and needed every dollar. I suspect the low-power architecture and a rock-bottom deal from Nvidia made the decision easy. Now that the future is bright, I doubt they'll repeat the mistake.�
Jul 24, 2015
HankLloydRight Hmm... I'm not convinced. BOSCH makes my dishwaher, but it's not very smart.
�
Jul 24, 2015
travwill Ha, true and same here. That was the Bosch vision, not anywhere near possible yet from their ability as a company. Completely made up and pieced together video, screens look mocked up and imposed we know that, and the whole thing was a little cheesy - i.e. rate this maneuver part. How do you rate a failure/crash, 0? so it does better after being repaired next time?
;-)�
Aug 3, 2015
TonyOrr Without giving the game away, could you give a few examples of what method of pestering they might be using? IE, tap something on the main display every few minutes, touch the steering wheel, hit a button on the wheel, push the TACC stalk, etc? Is it one of those?�
Aug 3, 2015
Stoneymonster You have to scream "I'M AWAKE! I'M AWAKE!" every 30 seconds.�
Aug 3, 2015
Footbag Yes, one of those :smile:�
Aug 3, 2015
Andyw2100 Well, this is interesting.
Assuming Footbag isn't kidding, I think we may be able to whittle the options down to tapping something on the main display every few minutes and pushing the TACC stalk. The steering wheel wouldn't have any way of knowing if it was just being touched, as I don't believe it has sensors for that. So I think we can safely rule out that option. As for hitting a button on the steering wheel, all the buttons on the steering wheel have functions that aren't going to be rendered unnecessary while TACC--even TACC with lane-keeping active--is functioning. So I think we can rule out buttons on the steering wheel.
That leaves the TACC stalk (which makes a lot of sense) and some sort of pop-up on the display (which makes less sense, since it involves taking your eyes off the road and at least one hand away from where it would be able to steer if needed.)
So my money is on something involving the TACC stalk.
TonyOrr specifically said "push the TACC stalk", but I'm going to guess that Footbag may have lumped anything involving the TACC stalk button into that as well. Since the button has no function other than to toggle TACC on and off, I'm going to go out on a limb and predict that the nag Footbag was told about involved the TACC stalk button. It could still be used to toggle the system on and off, but when being nagged, pressing the button doesn't toggle the system off, but rather answers the nag. Or Tesla could do something regarding the duration of the button push, just as they have with the duration of the pulls on the stalk.
So I say it's the button.�
Aug 3, 2015
Ingineer The steering wheel could definitely sense the mass of your hands on it through the torque sensor. When the wheel moves a little, there will be a larger torque sensed with your hand mass vs. just the wheel itself. This is also how the Mercedes system works, and note that it can be defeated by adding ANY reasonable mass to the wheel, it doesn't have to be a hand. (in the Mercedes case, taping a red bull to the wheel defeats it)�
Aug 3, 2015
Andyw2100 I saw that video of the can (or was it two cans?) taped to the wheel.
It just seems like if Tesla was going to go for just some acknowledgment that the driver wasn't asleep, they wouldn't necessarily want the driver interfering with the steering the car was doing. If we have to steer to show the car we're awake, doesn't that defeat the purpose?
Also I'm guessing if I saw that video the powers that be at Tesla did too. They're not going to want to design a nag system that could be bypassed that easily.
My money is still on the TACC button.�
Aug 3, 2015
donv A simple main display popup would be sufficient.
Boeing actually does something like this on the 747-400 and -8-- every 15 minutes, it puts up a warning message, which needs to be acknowledged.�
Aug 3, 2015
Ingineer My money is on no nags. We don't have them anywhere else, and Elon seems to be against them.�
Aug 3, 2015
loco I'm afraid a button on the main screen would be too distracting, because it would require to look away from the road ahead. In a 747 en-route on autopilot that is not a problem. Pilot is focused on instruments and nearest objects are usually miles away.
Hope we'll not be driving like Mercedes owners, but I can't think of anything other than steering wheel.
Maybe a steering wheel button?
�
Aug 3, 2015
spentan I sold my 2015 S550 for a P85D just so that I don't have to deal with this crap again. The S550 Active Lane Assist requires a torque on the wheel so distinctively, that sometimes I had my hands on the wheel and it wouldn't recognize it.
So I had to pull on the wheel, just to make it go out of line, which it would then correct. Totally unnecessary, and literally defeats the whole purpose of the system.
I'm really hoping that the nag can be disabled, with a huge liability warning etc, like the suspension lowering at high speeds.
I'm happy enough to stay awake without a nag, its just pointless otherwise.�
Aug 4, 2015
Andyw2100 I just want to be clear about what we (or at least I) was discussing.
When I said "My money is still on the TACC button" I was referring to the nag feature being tested, as explained to Footbag when he met the Tesla employee at the supercharger. I'm saying I think that's what is being tested, not necessarily that it is going to be implemented.
I certainly hope that we get lane-keeping without any sort of nagging. And yes, Musk is definitely on record as being opposed to nag features, from a safety standpoint. I'm not attempting to predict whether or not the first iteration of lane-keeping is going to come with a nag feature or not. I was merely providing my thoughts on what might currently be being tested, as it was described to Footbag.�
Aug 4, 2015
donv I assume you never adjust the nav system while you're driving then... or select a new station on Slacker or TuneIn... or any of those sort of things.
�
Aug 4, 2015
Andyw2100 I'll address that, because I had made the same point earlier.
The argument I'm making as to why Tesla wouldn't use a pop-up on the 17-inch display as a nag feature is that it makes no sense for them to --purposely cause a distraction-- and --purposely cause a driver to move a hand from where it can steer the car-- just to make sure the driver is still alert and not distracted, and able to steer the car if need be. I just don't believe Tesla would do that, as it's counter-productive.
Note too that currently no one is --required-- to use anything on the 17 inch display while driving. Tesla gives us the ability to distract ourselves if we so desire, but they are not going to be the ones to distract us, in order to make sure we're not distracted. It's not going to happen.�
Aug 4, 2015
TonyOrr I'm going with TACC stalk. Makes sense to tug it, push it, prod it or whatever once in a while if they need to. My preference would definitely be for a no Nag option! Having to pull the stalk to resume TACC after it's gone into 'HOLD' after 3 seconds is already a bind�
Aug 4, 2015
donv But... the point is that you're NOT driving at that moment, autopilot is. So looking down for a moment to dismiss a warning shouldn't make any difference. Oh, I suppose in the one in a billion case where the warning comes up just as the sudden fog layer descends or the deer jumps in front of the car, then yes, maybe it could make a difference. Or, the car could be hit by a meteor just as the warning comes up-- about as likely.
�
Aug 4, 2015
Andyw2100 You can bring up crazy and ridiculous scenarios to your heart's content. I'll stand by my position that Tesla is not going to require anything in the way of a nag acknowledgment as a method to prove that a driver is not distracted that in any way could be seen as itself being a distraction from the job at hand, which is operating the vehicle.�
Aug 4, 2015
Ingineer We can speculate all we want as a fun exercise, but we have to wait and see what they come up with. I'm interested in seeing how they deal with all the potential strange situations. Like what if the driver leaves the seat? Does the car stop in the lane? Keep driving? Pull over? If the system has to "disengage", and the driver doesn't take over (Sleeping. Heat attack, etc.) then what does the car do?�
Aug 4, 2015
TonyOrr I thought about this on my way home from work and have now changed my opinion. Using they TACC stalk is too close to the steering wheel. When I pull the lever stalk I use the steering wheel for leverage. Wouldn't be a good option if the moving the steering wheel disables auto-steering in any way so I therefore think the most likely option is a great big square on the 17" display with a countdown timer. Big enough to hit without being fiddly or too distracting. Ingineer, In the event of not hitting the screen in time then I'd expect the car to switch on the hazard lights and slow to a stop. Fun to speculate. Hopefully we'll all know 'soon'�
Aug 4, 2015
Andyw2100 What do you use for leverage when you press the button on the end of the TACC stalk?�
Aug 4, 2015
Ingineer I doubt touching the wheel will disable autosteer. It will attempt to correct when you do, but you will always win. Maybe after a prolonged control input it would, but it would be annoying if a little bump caused it to disengage every time.�
Aug 4, 2015
gizmoboy Automatically pulling over is awesome but seems fraught with edge cases (what if I'm on a bridge? No shoulder? Cliffside? Meteor is falling onto the shoulder as I pull over, etc.).
I love the concept of it being safer to have this enabled than not if a driver were to become incapacitated. But if it's not perfect, will people still bitch/sue/etc.?�
Aug 4, 2015
MarcG Bad scenario for Autopilot in a non-jammed freeway, which is the main use case for Autopilot.
Imagine this: you're in the fast lane going 65 mph, traffic flowing fine but fairly busy. You fall asleep or are too distracted to hit the "I'm still with you, please keep driving yourself" button in time. The last thing you'd want is for the car to stop on the freeway! Recipe for disastrous accidents.
- - - Updated - - -
Not possible with current hardware implementation (no rear radar or "proper" rear facing camera). Lane changes won't be fully-automated until adequate hardware is installed.�
Aug 4, 2015
gizmoboy That's a great point... all of the other TACCs I've used (BMW and Infiniti) will self-disengage if the conditions make it unusable.
If one assumes that a putative inattentive driver is considered another case for unsuitability, then disengaging is about the worst possible response the AP system could have.�
Aug 4, 2015
Ingineer I'm aware of the lack of rear radar and camera, and it's doubtful they could use the existing rear camera, but it may be possible. If incapacitated, I would rather it take "my" chances with a slow right lane change until the car can shoulder itself or find an exit ramp. If it begins signalling and looks for clear sonar, it should be pretty safe for it to slowly execute a lane change. Most drivers have had a human driver ahead of you change into your lane like this when you are going faster, I know I have on many occasions. Usually followed with a bunch of expletives, but no accident.
Stopping in the middle of the lane is super-dangerous, especially in hilly areas and/or with high speeds. Multi-car pile-up waiting to happen.
If it were me programming it, in the event of an unresponsive driver, I would have it first alert Tesla over the cellular link while maintaining the set speed. It would then begin by signalling and slowly getting right, (so slowly drivers will have time to react) while scanning for clear shoulder or an exit ramp, then pull over and stop with the hazard flashers on. Tesla could attempt to call the owner and alert emergency services.�
Aug 4, 2015
mikeash How about making some loud noises and flashing some interior lights first? If you're just distracted or asleep, that should wake you up. If you fail to respond, something worse has probably happened, and slowing to a stop (with hazards activated) is probably a decent reaction. That's probably a medical emergency, and being stopped with flashers will be better than letting you drive off the road, and ought to get bystanders to investigate.�
Aug 4, 2015
Ingineer TACC is different from full auto-pilot. Autopilot has a much better chance of lulling the driver to sleep since he isn't engaged. Regardless of the Legal issues, you want to do the safe thing, not the thing more likely to cause an accident.�
Aug 4, 2015
gizmoboy Unless you're just ?in flagrante delicto of an... intimate nature -- in which case you're probably getting a ticket.�
Aug 4, 2015
MarcG Tesla has yet to unleash a secret weapon for Autopilot that will solve all these issues. If the driver doesn't respond to a nag, an electric shock will be sent from a high-voltage harness that was secretly installed in the seat bottom cushion of all Autopilot-enabled models :biggrin:�
Aug 4, 2015
brianman Can I get that if I don't respond to the nag? Sounds like it might be good for my back.�
Aug 4, 2015
Nevek This makes sense to me, as pulling the TACC stalk is already the way (other than tapping the accelerator) to signal you are ready to resume cruise after a stop and the timeout. The same idea-"I'm paying attention, and want you to resume/maintain prior programming" would apply. Perhaps it would need the input after some change event being noted, such as extra freeway lanes, traffic cones, or a split freeway.
But I'd still prefer no nag.�
Aug 4, 2015
Footbag I loved the meteor comment!
Interesting comments by everyone, the can on the steering wheel 'hack' is pretty ingenious.�
Aug 4, 2015
MorrisonHiker I hope it doesn't nag too much...but if it does, maybe the entire steering wheel is touch sensitive? Then you wouldn't actually have to turn the wheel but just touch it lightly anywhere.�
Aug 4, 2015
Todd Burch There are a couple decent options here. I like the no nag option, but if a nag is required:
There's no need for the driver to do a specific thing to keep autopilot engaged. Think of a screensaver. ANY input to the computer will turn off the screensaver. I think ANY input could keep autopilot engaged.
1. Any change to ANY setting on the car (volume, window opening, tap on the touchscreen, turn signal, TACC distance adjustment, pedal input, etc.) could reset a nag timer, because that means the driver is conscious. Suppose that nag timer is 3 minutes (a reasonable value IMHO). If you make an adjustment to SOMETHING within every 3 minutes, you'd see no nags at all. That will get you a long way, I think.
2. After three minutes of no control change of any kind, a little message appears on the dash and prompts you if you want to continue autopilot. Again, ANY input to the car resets the timer, since it indicates the driver (or someone else in the car) is conscious.
3. If the message is not responded to within 30 seconds, Music fades out (if playing), a gentle tone sounds, and the car again waits for any input. If there is no input after 15 seconds, a loud, annoying alarm tone plays for perhaps 5 seconds in an attempt to wake the driver. If there is still no input of any kind, one must assume an incapacitated driver, in which case the car should make an attempt to pull over safely. (I think it may be possible to do this with current hardware using hazard lights, the camera, sonar, and very slow lane changes). The car would then prompt Tesla to call you, and if there is no response, contact 911 and send an ambulance to your car.
This way, if a nag is required, it starts out in the least annoying manner and gets progressively more annoying as the concern that the driver is asleep or incapacitated grows.
This would be a good balance to minimize annoyance but still have some sort of confirmation for safety reasons.
You can't eliminate every danger or corner case, but I think these steps would be reasonable and fairly non-invasive.�
Aug 4, 2015
Andyw2100 While I personally like the idea of "any input", from what I've heard about our cars' functions, I don't think that's likely to happen.
The people that have some insight into how our cars' software is designed seem to be fairly certain that the driving and safety software is completely and totally separate from the other stuff, like the audio player, navigation system, etc. According to them, there are good reasons for this separation. Your plan would require integration of the two separate parts of the software. I don't think that is likely to happen, if it is even possible.�
Aug 4, 2015
Todd Burch Sure it's possible, and I think very doable in fact. It's true the vehicle functions must operate without the other systems (so you can drive while rebooting for instance), but that doesn't mean that the vehicle functions cannot respond to input from, for example, the touchscreen. The touchscreen can control creep, traction control, the parking brakes, and other systems, for instance. No reason to think the autopilot system couldn't respond to inputs from other sources. In fact, I think it's 99% likely that what I described is perfectly doable.
The key is that vehicle systems must operate independently of the other systems. But that doesn't mean there can't be communication between them.�
Aug 4, 2015
gizmoboy 3 minutes! :scared:�
Aug 4, 2015
Andyw2100 OK, I hope you are correct, because I like that concept.
I probably misunderstood what others had written about the way they believed the software was probably handled.�
Aug 5, 2015
gizmoboy Thing is, if they want assurance that you're paying attention to the road, having you check media or change temperatures is no assurance (in fact, it's assurance you're not currently paying full attention to the road).�
Aug 5, 2015
MarcG One interesting tidbit from the shareholder call that is currently ongoing: Elon said that the "nag" feature will highly depend on how early access customers will respond to it.
And by the way, he expects the auto-steer and self-parking features of Autopilot to be release to early access customers on August 15th (in 10 days).�
Aug 5, 2015
TonyOrr Yes, so we need to ensure all the beta testers insist on no (or minimal) nag :smile:�
Aug 5, 2015
Todd Burch It is not Tesla's responsibility to ensure that you're paying attention to the road--that's the driver's responsibility--and it would be darn near impossible to do that. I don't believe that's their goal at all.
The nags are about ensuring that the driver is awake/conscious.�
Aug 5, 2015
flowbee Better prepare to wait a while longer for Autopilot..
From today's investor call:
"Tesla is targeting release of its software update to beta customers this month, and depending on how it is received, a wide release for the auto-pilot and auto-park features in one to two months after that, Musk says."
Source: http://blogs.marketwatch.com/thetell/2015/08/05/live-blog-tesla-to-report-second-quarter-results/
Assuming the early access release is Aug 15, this puts general Autopilot release in the Sep-15 to Oct-15 timeframe -- assuming everything goes well and no major issues are encountered.�
Aug 5, 2015
AlMc I think the general release coincides well with the start of delivery of the X which will certainly have as many, if not more, sensors for AD. I suspect having it ready for the X launch may be the ultimate goal (and time restriction).�
Aug 25, 2015
AziwA Hi Footbag.
With the recent "leaks" of v7.0 on this forum. Can you comment on any of the differences between what you saw and the pics we have all seen?
Thanks!�
Aug 25, 2015
Footbag The 17" screen looks just like what I saw. In short, new coloring and some small design changes, but essentially the same as what we have now. As the car I saw was off (brake pedal not pressed), I didn't see much on the instrument panel. The one change is with the energy graph. What I saw was more stylized, with a bubble/round/3d look. The leaked graph seems basically the same as our current one, and I did not see the power meter around it, as in the leaks. I am not sure if it has been confirmed if the power meter shows when the car is off in the leaks?
The lock button I saw is in the leaks (top left of 17"). The one Tesla employee did comment that you would now be able to see brake lights via the center of the instrument panel, so that seems to be the case. The other, main source of info indeed commented that the center portion was significantly redesigned and that it would look pretty cool seems to hold up... when autopilot info is shown. I would tend to agree with WK057 (going from memory, sorry if I got the name wrong), that they have reduced 'design/fancy looks' to speed the system up.
I plan to revisit my first post once all this washes out, if only to see for myself how close/far it ended up being from the final product.�
Aug 26, 2015
AmpedRealtor Apple didn't go to a flat UI look to speed up the graphics, and neither did Tesla.�
Aug 26, 2015
gizmoboy In your (outsider) opinion.�
Aug 26, 2015
AmpedRealtor Of course it's in my opinion, why would I be posting someone else's opinion? lol
And just to remind you, we are all outsiders here according to your definition.�
Aug 26, 2015
HankLloydRight The first iPhone in 2007 had 3D graphics and full-on skeuomorphism, and it ran just fine.�
Aug 26, 2015
AnOutsider ahem...�
Aug 26, 2015
gizmoboy In case that was too subtle for you, I was pointing out that you stated something as simple fact about which you have zero substantiated evidence.
And yes, we are all outsiders of Tesla (AFAIK), so we might want to take care when stating "facts" about their motivations.
- - - Updated - - -
I agree with AR in part: Apple's move away from Skeuomorpism is documented to have nothing to do with performance and everything to do with style and taste.�
Aug 26, 2015
gg_got_a_tesla Well, you are AnOutsider, not TheOutsider...�
Aug 27, 2015
AziwA Very interesting Footbag, thanks!�
Aug 27, 2015
AnOutsider brb, hacking username table.�
Aug 27, 2015
AmpedRealtor Not subtle at all, but thank you for the condescending follow-up. And in case my point was too subtle for you, everything anyone says here is, by default, opinion. Perhaps the only real facts ever posted were posted by George Blankenship because he actually worked for Tesla and had the facts. Mine happens to be an informed opinion based upon my understanding of how GPUs function. I worked in graphics and UI design for 10 years and have a pretty good understanding of why these types of changes are made. I wasn't always a Realtor.
Suggesting the reason is due to performance demonstrates a lack of understanding of how these systems function.�
Aug 30, 2015
ohmman I haven't seen much chat around here regarding Elon's tweet that all cars would learn from other cars. This implies a supervised learning algorithm of some sort, which digests collected data from cars on the road (I assume that would include geolocation and other important metrics). Does anyone know if there any other car manufacturers at this point who are collecting this kind of information for the purposes of supervised learning/steering control, etc?
I would guess that this only applies to Autopilot vehicles, but certainly an argument could be made for valuable data even from 'classic' vehicles like my P85. However, assuming it's only Autopilot vehicles, the input set is limited by those vehicles and driven miles. If another manufacturer is doing this, I would guess they'd catch up really quickly because they have the ability to use a much larger set of vehicles on the road.
It would be really interesting to get more details on this program from Tesla, though I can't see much benefit in making it public. I imagine a map with Autopilot confidence levels highlighting the roadways.. would be cool.�
Oct 15, 2015
Footbag With v7/Autopilot now released, I wanted to revisit my original post to see how things turned out comparatively:
- Autopilot is very close to ready - i did not press for more specifics. They were unaware of Elons comments about the 405 or whatever highway with faded white lines and light concrete.
Well, it was about 2.5 months later that it got released. So apparently it wasn�t that close to ready. Perhaps Elon decided to hold things up while they worked on faded lines and the like�
- To that end, there is consideration that the system may talk to the mothership to know areas where autopilot can't be relied upon, so the car will give the driver notice in advance - "you are approaching an area where autopilot cannot function reliably, prepare to take control" (my words based on this persons comments)
Yep, Elon went into some detail about how this crowd sourced data is going to benefit all Tesla drivers. It seems several instances of the �nag� that people (including myself) have faced could in fact be the �autopilot cannot function reliably� message in another form (albeit it is too soon for it to be because of lessons learned from other drivers).
- Having seen a situation where a 110km speed sign was read as 10k by the car, I mentioned this to this person. They were quite shocked this happened. They did confirm the intention is that the system will automatically adjust the speed of the car (as seen in the initial demonstration, and as an improvement to the current "set cruise to speed" function we have now).
Since we don�t have the speed auto adjusting, I have to believe they have seen issues with speed sign recognition, as I mentioned. It should be noted that Tesla has been improving the speed sign recognition, there are confirmed cases where a sign that used to cause issues now registers correctly.
- I suggested/asked if the system had a threshold - don't adjust speed if delta is greater than x. That is not part of the system, but they were intrigued by the idea.
Since they don�t adjust speed, this is still a wait-and-see item.
- I asked if the car 'tells' tesla when a speed sign is 'ignored' by the driver, ie that the car detected it wrong. No such communication happens.
Similar to last two points, too early to know much. With Elon�s recent statements about the cars sending back info, it is reasonable to believe this may be included, but that is just speculation.
- In stop and go traffic, the system is reliable enough that the driver could be distracted while driving and still be safe (not saying this is a good thing or the intention!). This individual gave examples of doing tasks very not related to driving, and having nothing untoward result.
So far, every indication is that this is correct. Several reports of reasonably long distance travel without a nag or incident supports that in certain situations, the car is more �autonomous� than not. Of course, since situations can change in an instant, drivers still need to stay alert.
- There appears to be a bit of 'debate' within Tesla over how much the car should pester the driver to stay focused on driving. A 'new' feature just put into the software recently goes quite far towards pestering. This individual really disliked it, and was trying to get it reverted. It has to do with, and is in stark contrast to some statements Elon made a while back. I joked "the lawyers must be getting involved" and the response I got was a slight chuckle. We all will hope this individual gets this change reverted (but I am not so sure Tesla as a company will revert it)
As above, in many instances, there is no nag. Often times, the nag is less a nag and more a notice that intervention is required for safe travelling. However, I have experienced what I would call a nag� taking a long curve (not too tight, but length) on a highway, with cars around, AFTER completing the curve, the car twice (once in each direction) asked to grab the wheel. The second time I refused. After a few warnings (about 30 seconds apart), the car slowed down. However, autosteer remained on, and TACC remained active (as indicated by the icon lit in the dash). So long as the accelerator was used to maintain speed, I was able to keep driving hands free. However, after about 2 minutes of constant beeping (rather annoying beeping), I have in and moved the wheel slightly. This was all it took to make the car happy, and TACC took over again. Since it went on for 2 minutes, and since the car never showed any signs of having difficulty, I consider this a nag. Granted, I was then able to drive nag free for ~20 minutes.
- I brought up Elon mentioning sensor suite 2. This person confirmed that more sensors are needed for autonomous driving. No surprise. Did not ask about the Model S seen with additional cameras.
Elon�s comments yesterday about the coming ability to recognize traffic lights and stop signs was ambiguous, but it seemed like it applied to the current sensor suite, so there could still be improvements to be seen with the original sensor suite. He did reaffirm that more sensors would be needed for autonomous driving.
- Some discussion was had about laws - Tesla is keenly aware that autonomous features in cars are a grey area, that laws are typically not in place about them, and Tesla wants to implement it to be safe enough that laws don't need to get written, with the fear being that when government writes such laws, they will be too overboard on safety and will prevent features.
There have been suggestions that even the current abilities of the car are running into issues with the law. I can�t recount the source, but there was something published that the car will behave differently in regions so as to comply with their road rules. Also, Elon stated the rollout of v7 to Europe was going to be a smidge later, as they needed some regulatory approval.
- I mentioned having owned an Infiniti Q50 for a short time - "oh, the car that ping pongs between the lanes"... this person knew about the car. It was mentioned the team has a Merc S Class, and that is the benchmark they are trying to beat. There was confidence implied that they had beat the S Class in lane keeping.
Elons response to why Tesla�s autopilot was better than the competitors was answered mostly with a reference to the ability the cars have to communicate with each other (in a sense). No competitor has anything like that ability.
- No discussion about self parking - I mean, this person got out of the car and it parked itself next to the Supercharger, but I wasn't about to point out that we all saw that (just kidding)
Well we got parallel parking. Apparently more abilities are coming in v7.1 (with no set timeframe given yet0.
- This person mentioned it was much more relaxing driving in autopilot mode. Something said to the effect of 'if you are tired, you still are able to monitor the car just fine (autopilot on), whereas you might not be as able to drive the car safely (autopilot off)'. Again, I don't think this individual nor tesla (nor I) advocate driving a car with autopilot when you shouldn't be/when you wouldn't be able to safely drive a car without autopilot.
Driving on the highway/freeway in AP is certainly a pleasure. Driving on random city roads are hit and miss, but there absolutely are situations where driver intervention is needed in a hurry.
- I asked if I could see v7 which was running in the car, after a quick 'thought' ('yes' was not immediate), I was let in the drivers seat - it was made VERY clear not to try to take any pictures. I was very surprised to be allowed to sit down.
- The person did almost all of the interaction, except when I asked to press CONTROLS once (this was out of respect, but it seemed like this person would have been ok with me pressing buttons at will)
- I didn't think to press the T to see version number
- As talked about on here (TMC), it is the dark/flat look that was in the demo cars at the D launch.
This is indeed the case.
- I was a bit overwhelmed, and didn't take too many mental notes about the specific looks. Icons were less cartooney.
This is indeed the case.
- There is less contrast between the background and the text/icons - sun was shining brightly on the screen, and even the individual had trouble making out the bottom row of buttons (settings, temp, etc).
This is indeed the case. That said, I didn�t have issues with visibility thus far with the sun. Contrast between elements on the screen may have been tweaked from what I saw (tough to recall).
- The layout exactly mimics what we have - everything is in same place on the 17" screen
Indeed the case.
- There is a new icon indicating charge port is open
I actually haven�t checked this myself yet.
- On a very quick pass through, the CONTROLS screen seemed to be the same. The individual jumped through some pages, I didn't notice anything, but certainly may have missed items.
- It was implied that there were not any BIG new features - I did not see an APPS icon for instance - and it is still the same 2 open windows layout. This relates to my perception the person would have let me press buttons at will - in that there wasn't anything present they thought needed to be hidden.
Aside from the clock and individual tire pressure read outs, no big new features came about (yes, some features were removed, much to the dismay of several forum members).
- The top icons (nav, music, etc) did not seem to be 'in a tray' as they are now (or so my memory recalls - again, didn't think to pay enough attention)
Indeed the case.
- This individual stated being very bothered when various button presses would take time to activate - so the team spent considerable effort speeding response times up - but are limited by the hardware (I did not ask about model X having a faster processor)
I haven�t noticed much in this regard (for better or worse), but WK057 certainly things it is faster, so perhaps this individuals focus on performance bore fruit.
- This person never uses the web browser (who does LOL), so couldn't comment if it was much faster. When launched, it was set to go to youtube.com - but didn't load in a few seconds and we moved on. Typing this now, youtube is an odd site for it to automatically launch to, I wonder if they added in video playback? but again, the site didn't load, so that is just a guess on my part.
I haven�t thought to check� but I would be shocked if video playback was added.
- Car was running 3G - From what I read here, LTE hasn't really sped things up for owners, but this individual indicated it should, FWIW
- Nothing obvious changed with backup camera.
- Panning in Nav seemed smoother/quicker.
- Did not see Music or Energy windows.
- I did joke about there being a bunch of little features that likely were added, the person indicated indeed there are.
This seems to be the case. Tire pressure read outs, the clock (yippie?), changes to trip metering, shortcut to lock/unlock doors.
- One such little feature, and one this person never even noticed before I pointed to it (in this persons defence, autopilot was their focus, not the V7 overhaul), was a LOCK icon at the very top left of the screen. It appeared to indicate if the doors are locked or not. This individual never actually noticed it there before. Did not test, but I suspect it may be a quick way to lock/unlock the doors - yay.
- The instrument cluster follows the same dark/flat approach.
Indeed.
- The energy graph and nav were on screen. As much as I said 'flat', these two were 'rounded', in the sense of round dials - the energy graph no longer is the rectangular design.
So the energy graph changed substantially from what I saw. The NAV I think is quite close� I like how the new NAV fades into the background.
- v7 UI Overhaul is very close, again except for the rounded energy graph (and I'm pretty sure Nav was rounded as well, but less certain)
- As the car wasn't in motion/on (i didn't step on the brake), nothing of note was displayed in the center area.
Well we all know what is there now!
- It was mentioned that the center area got a large redesign. I honestly couldn't grasp/picture what was being explained to me - it sounded like the round dial interface was gone - which would seem to be represented in the above link/picture.
- The center area is going to show the cars position in relation to the lane markings
- It (or somewhere) will also indicate the position and distance of cars/objects around you - "very dynamic" were the words used.
- It was mentioned that if you got too close to an object beside you, the car would move you away. I interpreted that as a means to prevent/reduce impact of side swiping.
The above 4 points all turned out to be accurate.
All in all, the info I got seemed to be pretty darn accurate. Lesson learned... it is well worth talking to Tesla employees at Superchargers!
�
Oct 16, 2015
beeeerock I had the opportunity to drive from Kamloops to Vancouver this morning, over the Coquihalla, with the new V7 firmware just loaded into the car. So 300 to 350 km of divided highway driving, mountainous terrain and through the flats of the Fraser Valley. I don't have any real differences of opinion with your comments, but I did notice a few things that might be of interest...
- It doesn't take much user interaction to cause Autopilot to release. I found that if I was a little nervous about how close a car or the median concrete was to me and just slightly applied pressure to the steering wheel, it would beep-boop off line and I'd be driving on my own again. Unless some other factor was contributing, I found this to be a little too sensitive. It would be nice if it functioned a little more like cruise control... where you can add some speed with your right foot for a period, then let go and have your speed drop back to cruise speed. Instead, it acts like you touched the brake and it just releases. I'd like to be able to add my own inputs, then let it take over when I release the pressure.
- Because of (1) above, I found I didn't want to hold onto the wheel, or allow it to take the weight of my arms. My shoulders got tired from holding my arms up so they didn't interfere with the Autopilot and cause it to drop out. I ended up having them palm up on my legs, just under the wheel so I could grab when needed. The reminders and disclaimers indicating you should hold onto the wheel seem to disagree with the reality of how holding on prevents the system from working.
- It gets confused when a climbing lane is introduced on the right side of the road. I found it would follow the fog line as the road widened, then suddenly 'realize' that the new lane line was there (about to straddle it) and swerve back to the left and into the original 'through alignment'. That was weird and a little disconcerting.
- I also got the 'nag' in places I didn't expect it - at least based on visual information available. Long corners were the usual, but a bridge abutment also freaked it out.
- Probably a good thing - as I approached a construction zone, with orange delineators all over the place, Autopilot decided it wasn't going to take on the responsibility of keeping me in my lane and told me to take over NOW! I was ready for that, since I was pushing the envelope to see what it would handle. This was a circumstance that it probably shouldn't be allowed to control - yet.
- The new dash shows the current speed zone with a 'sign', rather than the bar on the speedo. Much more intuitive. I had very poor luck with my camera recognizing speed signs in 6.2, so I was hoping for improvements. Not the case. Now that the speed limit is more obviously displayed, I was more aware of how often it's wrong. And that turned out to be worse than I thought in 6.2. It never got the speed right on the Coquihalla. In spite of being posted at 120 on big signs, the dash told me the limit was either 90, 100 or 110... but never 120.
- Driving on the long flat straights in the Valley was really nice. Autopilot behaved quite well throughout that whole run and I didn't feel like I needed to have my hands *right there* at all times. Good or bad, I don't know.
- I really liked that it didn't pinball me between the lane lines like the Audi 'lane assist' function does!
- As far as the general user interface goes, I do miss the energy indications.
- I might get used to the different information displayed in the trip meter, but it didn't feel right to me today.
- I like being able to check tire pressures, but not sure they should be where they are... a page in the settings somewhere might be more logical.
- The night mode didn't want to flip to day mode as early as I thought it should (yeah, out the door Way Too Early). I changed it manually rather than increase the dim night mode intensity.
Overall, I think Autopilot works quite well. Considering it's still noted as 'beta' in the settings screen, it feels really good. Like many others, I think that the distinction between Autopilot and Self-Drive is a bit blurry. I know what it means, but to operate as I *think* it was intended, I believe it should allow more input from the driver so it's more of an assistance system. In other words, I should be able to hold onto the wheel and have it keep me between the lines, but not bale on me if I want to avoid a pothole or something... I should be able to interject a swerve and have it pick up again seamlessly.�
Oct 16, 2015
Andyw2100 Great post! Thanks!
I can see how and why this would be useful, but I can also see why Tesla wants to just "yield control" if the driver is attempting to steer. It is really simple to give steering control back to the car, which I think makes this less of an issue than it otherwise might be.�
Oct 16, 2015
Cosmacelf Hmmm, just like there is a simple tap up and down on the TACC control To make 1 mph adjustments, maybe there should be a way to tell the car to align itself more left or right in the lane.�
Oct 17, 2015
Soolim It might come in the final version, if enough feedback / interest is made by the user i.e. TMC .�
Oct 17, 2015
beeeerock I agree, there is definitely some logic to how it works right now. But in my case - and perhaps on the learning curve - there were a couple of times when I didn't realize my hand had been heavy enough to trip Autopilot offline and had I not heard the double beep, might have released my hand completely and ground up against a concrete median barrier on a long curve. If I don't leave the lane with my 'intervention', I'd like Autopilot to remain on. Miss a bit of blown out tire or a pothole etc...
Another example was passing a wide load on a sweeper curve with concrete no-post on the left. Autopilot seemed to know where the barrier and the truck were, but didn't completely calculate the overhang of the excavator on the back. There was space, but Autopilot wanted to put me down the centre of the lane, which felt uncomfortably close to the load. I blinked before Autopilot did and adjusted the steering closer to the barrier. It tripped out. And left me to carry on. I was ready, because I was trying to find the limits of the system, but I didn't feel comfortable with how it managed that situation.
It just feels like a little more fuzziness in how it decides to trip offline would improve the situation. It could even be a setting like with regen - normal or minimal - just something to determine how little intervention will shut it down.
Somewhat related - I wish that TACC wouldn't leave the braking to the last minute. I've been hoping this behaviour would change with ver 7, but I still hear the brake mechanism activating and feel myself pucker as we approach a slower or stopping car! This will NOT work well if the roads are at all icy or slippery. Yes, TACC perhaps shouldn't be used in those conditions, but people WILL use it and it's going to cost in insurance claims.�
Oct 17, 2015
511keV I found my Model S continually "yielding control" a bit annoying when traveling in the right lane on highways when the car wanted to take every exit and I tried to guide the car to continue on the highway. Hopefully, all of the "learning" that occurs whenever a driver takes control will fix this over time. Given that I'm far from the mothership and probably where most beta testing was done, the highways exits around Boston probably aren't as well known to the navigation subsystem of Autopilot (Beta,) so I learning to live with it and, hopefully, watch my feedback improve things.�
Oct 17, 2015
wcalvin Maybe just a long enough manual steering override to set it to the new lane position for the next 30 sec.�
Oct 17, 2015
Andyw2100 I actually started a thread on this a few days ago. It's here:
Possible Autosteer Enhancement - Temporary Left or Right Side Lane Preference�
Oct 24, 2015
Panu Looking at the videos from USA it looks like AP changes lane after tapping turn signal without requiring the driver to touch the steering wheel. Today Tesla owners have received v7 in Finland and they are reporting that you have to initiate the lane change by 1) tapping turn signal and 2) turning steering wheel. Is this behaviour EU only or did they already change it to require more driver input? The version they have in Finland is 2.7.77.�
Oct 24, 2015
Andyw2100 That is a later version than the version released in the US.
The US version currently does not require us to touch the steering wheel to initiate a lane change.�
Oct 24, 2015
ecarfan The way out works in the US is as you describe. No steering input is required. Sounds like the way it works in Finland is different. Interesting.�
Oct 24, 2015
Soolim I like the two step process for lane change, while a quick tap on the turn signal will allow the driver to provide left or right bias while lane keeping.�
Oct 24, 2015
Panu Yes, it's interesting to see if it remains the same in US in version 2.7.77 and above.�
Oct 24, 2015
beeeerock I wonder if this has anything to do with the delay in release outside of North America.�
Oct 25, 2015
Dborn Lane change in Australia requires you hand on the wheel, not initiating the turn, rather telling the car " I'm here". Then turning on the turn signal until the dotted lane marker becomes solid.�
Oct 25, 2015
Matias Same in Finland with 7.77.�
Oct 25, 2015
Andyw2100 I understand what you are saying, but don't understand the logic behind it, as it would seem being able to move the directional signal should be enough of an indication to the car that a driver is present and paying attention.
I wonder if the need to touch the wheel is meeting some sort of regulatory requirement.�
Oct 25, 2015
Soolim I wonder what sort of regulatory requirement Japan will impose for lane change safety.:wink: Not aware of any Japanese car having this feature yet. Anyone knows?
- - - Updated - - -
Have you tried it, and do you feel safe using it?�
Oct 27, 2015
Cebe In some of the countries, the rule is:
1. Check if it's clear
2. Signal
3. Check again
4. Change the lane
By letting signaling (step 2) initiate the actual lane change (step 4), you're missing a step.
Not defending the rule, just sayin'.�
Oct 27, 2015
beeeerock I think there is a big danger with people assuming the AP software is more capable than it is! Or, with people assuming it is *intended* to do more than it is.
I know there are plenty who will disagree with me, but I don't think drivers should be driving with their hands free... texting, eating, whatever... while AP is left to do the work of driving.
Driving on freeways in stop and go traffic might make this behaviour almost reasonable, but the bottom line is that the driver is responsible for what his car does. I've learned that AP can make decisions very quickly and cause the car to do things suddenly and almost without warning. That's fine with me - it IS beta software after all. As responsible drivers and Tesla owners, we want the company to succeed and our actions should support that goal. IMHO that includes ensuring AP doesn't become a huge mistake, by using it responsibly. We shouldn't be risking our lives or the lives of others in our path by 'pushing the envelope' carelessly.
The software pops up a notice when activated that the driver should continue to hold the wheel. I don't think that's just a legal disclaimer to be used as a defense by Tesla if something goes wrong.
I'm treating AP as a tool to look after keeping the car centered in the lane and help me keep track of other cars around me, both in front and beside. I'm driving with it on AND with my hands on the wheel. If and when it does something unexpected, I'm able to catch it before the car has tracked more than a foot or two off my desired line. AP looks after fine-tuning my driving and letting me not concentrate *quite* as hard. Not go to sleep or catch up on my emails.�
Oct 28, 2015
Soolim Voice of reason. Spoken by a driver who has tested the auto-steer extensively.�
Oct 28, 2015
garygid I believe that the implied delay between 2 and 4, intended to give fast-approaching
(and other) vehicles a suitable warning of your intended lane change, is a critical
part of improving the safety of lane changes.
Using the turn signal to indicate only "I am in the process of changing lanes" is not
enough, since most can see that your car is moving sideways. Using it to signal
your intention to move into the adjacent lane, BEFORE you start the maneuver, to
give others ample and sufficient warning, should be the major goal, for safer driving.
So, the "delay" (implied in step 3) with the signal going is a very important feature
for a safer AP to implement.
- - - Updated - - -
Is there any real indication that there is any way in AP at this point to set a preference for
following the left or right lane-edge marking?
If so, that would help AP a great deal, since on most all lane splits, the AP has no
idea which lane we want to enter, and we currently need to take over steering
(which cancells AP completely) to enter our own lane of choice, which might
be different than what we did yesterday.�
Oct 28, 2015
Soolim My preference is: Driver does step 1, 2, 3, and then turn the wheel for AP to implement step 4.
- - - Updated - - -
I fully support this implementation of AP auto-steer in future AP release. I feel safer using it in that manner.�
Oct 28, 2015
beeeerock Hmmm, just when I thought the car was learning to behave at a few spots on my regular highway, it's taken to (apparently) randomly deciding whether to do what I've 'taught' it and just doing what it did wrong. I've tried setting the NAV to imply a turn would be wrong and I've seen it randomly misbehave when following another vehicle that follows the alignment properly. So I'm not sure how much learning is going on and how much decision making depends on weather, surrounding traffic, speed etc. Or if it's just Beta software issues that will be improved as time goes on.
My perception is that it gets it right about 2/3 of the time now. And with time between trips, I almost wonder if the learned behavior has a Time To Live value connected to it... which causes it to expire and require re-learning.
It sure would be nice to have access to a white paper of some sort describing more of the details of how it works. I can't help but think there are Tesla folks in the know reading these threads and betting with each other on who will be the first to reverse-engineer the finer details of the AP system, simply by testing and watching how it behaves...�
Oct 28, 2015
ohmman I mentioned this in the other AP (improving already) thread, but I'm headed to NIPS in December. I plan to actively hunt Tesla's contingent, if they are there. If they're not, I'll be a little disappointed as it's sort of the king of machine learning conferences. Hopefully I can get a little more data on how it works and at least translate that to the forum afterwards.
Speaking of NIPS, anyone else in this forum going? I'm hoping to take one of those black Tesla taxis from the airport in Montreal...�
Oct 28, 2015
beeeerock That would be great and I'll look forward to your follow-up!�
Dec 14, 2015
ohmman I'm back from NIPS. Great conference, and very well attended. Sadly, I never saw a single person from Tesla or Mobileye. Everyone was wearing a badge with their institution, and I did scan aggressively. That said, there were 3700+ people there. I can't say I saw them all.
On the bright side, lots of companies and institutions are working on this problem. Andrew Ng talked quite a bit about their work at Baidu on solving it. Also, quite a few people were doing this work for autonomous mining vehicles.
Sorry I couldn't provide better info.�
Dec 14, 2015
beeeerock Thanks for the follow-up anyway!�
Không có nhận xét nào:
Đăng nhận xét