SageBrush
REJECT Fascism
I'm probably not understanding the changes, but I keep thinking that AP that turns itself off without user agreement is itself going to cause accidents.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Tesla will soon introduce new Autopilot safety restrictions after recent accidents
- Thread starter Tam
- Start date
I think the issue with alerts escalating is that the alerts are relatively quiet in 7.1. There's a tiny visual, then an audible shortly before a loud audible. I'd like the visual to be more, well, visible. My AP doesn't sense my hands on the wheel pretty often, which I think is a specific issue for my X. It's at the SC for other issues and they're taking a look at that as well. I have to wrestle with the wheel quite a bit to make it realize I'm still there. So in my case, yes - alerts can quickly escalate. I'm hopeful that 8.0 (or whatever version contains this update) will be more aggressive about alerting me before it disables. If so, I'm completely supportive. Not onerous on me, but limits inattentiveness elsewhere.
This has been my experience also. Then again, I mostly keep my hands on the wheel and my eyes on the road. I'm not really sure what everyone is trying to achieve that they are so upset about the nagging. We have already seen that it is a bad idea to do fun things like clean the windshield while AP is engaged. I mostly check google maps on my phone and compare it to the Tesla map while AP is engaged. otherwise my hands are on the wheel. i'm done almost 3,000 problem free road trip miles using 7.1 AP and think its awesome.
Congrats, you just showed yourself to be ignorant three ways in a single post! Hijacking a thread to make a political point, willfully misunderstanding the person you quoted, and disparaging a very smart guy.
Ulmo
Active Member
Trump is a boaster. These days, there's so many easy quick fixes available, it's embarrassing; our politicians haven't been doing the easy stuff to fix our country, let alone the hard stuff. It's crazy. It should never be this way. That's why Trump thinks he can do easy stuff and get away with it: he's half right about that, and we should have never gotten in this position. The normal situation should be that things are relatively difficult to do; life is difficult as a norm. So, even if Trump does some of the easy good stuff, right away, things will start working right, and suddenly everything starts to get back to normal difficult again, and the easy stuff alone won't cut it; it's only the stuff we've been missing that we can fix right away, but then we have to keep going with the harder stuff.
But, let's focus on wk057 & AP: I've been very consistent in saying that one person can program a good automatic driving program, whether it be Hoyt, wk057, or whomever is smart enough. (1) there's no one else in the team to get in the way and (2) management disputes can be resolved without having to compromise with people that don't have the same brain as you (since, after all, you're just doing all this with yourself), and (3) you're not distracted with the nuisances of other people in (1) and (2) all the time. You see an issue, and bam, solved in 10 - 200 minutes, whatever it takes, not making a committee or going to their cubicle for a powow.
Having said all that, the big corporate structure, if done with enough vision and innovation, can also work. As far as I'm concerned, each type of group (single or large) has the potential to succeed or fail, and much of those probabilities due to their respective advantages and disadvantages. If anything, a group of properly managed Hoyts and wk057's would be more certain to get it done than any of the other above combinations. But, we've already seen how that never came to pass with the specified example namesakes, and I already have a decent feeling of a part of the reason for that. For instance, in my case, I can't get any job that requires a Resume, so the Human Resources Department at Tesla would never look at me, and I don't have $1M spare to play around like wk057 or a lucky run with the college system that allows me to work at Google and other places like Hoyt. I have a feeling the types of actions it takes to succeed at business that gets a Musk to where an Elon is doesn't allow him to easily think in terms of making the types of compromises necessary to come to terms with such team members, and already we've seen words to those effects from both parties. It just is. Where we go from here can basically be nothing other than exciting. I'm looking forward to it, although it might be (already is) messy along the way.
Last edited:
I don't see how you can have messed with 8.0, it's not even out and we don't even know what they have really changed, for all we know they could have changed the algorithm completely, how can it be disappointing?
You are disappointed with tesla ( and for me this is fine ) but it's silly saying "the next it's a cover-up for adding nagging"
they said it will be better in the 8.0 and that there will be less nag ( but not because they lowered the level, but because it will behave better - this is my reading of course but it's the only logical reading.. )
now.. that's hard to belive unless the software/paramters/hard work it's already done by tesla, so you are telling us that the system it's already good they just too pointed on the 'safe' side? i would say that they surely want to put out what you have, but it's clear that the world is full of stupid people and they are forced to err on the safe-side, but again.. we were said many times lately that a huge improvement it's already done and they are just waiting to be 'perfect for the public' to announce it, i would not be so pessimist as you, but then, i don't have a tesla and maybe i didn't wait years with all my hopes broken every time, so maybe i can understand you
about that.. i completely agree with you, for a programmer ( and i'm assuming you are one like me ) seeing all the thing that can be done and left undone.. yesterday i was thinking of a lot of automation that can be done with the geo-referencing and in my mind i was thinking 'it will never be done..', it's heart-breaking.
i hope that when they are more free they will really put some effort on this side like with the radio etc, i would say that in 1year of work with just 4-5 people they could make a so-beautifull system that will gain them so much on this side.. just thinking of how cheap is to do this compared to all of what they've done..
Ulmo
Active Member
Your definition of "out" doesn't have the effect you think it does where wk057 is concerned.
I'm not certain of the particulars, and it doesn't really even matter, but from what I've heard, there are alpha testers, beta testers, and some inside work duties that wk057 himself has had from time to time, and all of that allows the theory that potentially various limited and controlled exposure to wk057 of potential future versions can and does happen.
I have a fairly high level of confidence that wk057 is a good actor, and furthermore that Tesla will eventually figure out a fairly safe autonomous car and various "autopilots" along the way to that path. (I just don't know how or when, and I think that's the exciting and interesting part for right now.)
Last edited:
Please, explain better.
The only way he can know something about it, it's to get the beta download ( and it could be really hard, or relatively easy based on how it's made the ota update procedure ), so, have him got the latest?
Wk057 previously posted screenshots of the flags. Everything he spoke about existing firmware is truth. I can't comment on 8.0.
I didn't doubt about what he has done or about the existing firmware, read better what i wrote, i said that i doubt he can improve the autopilot if it wasn't already good but just down-powered. without the source code it's really hard do manipulate the car to this level, what is doable is to 'adjust' some paramters ( this is related to the autopilot )
About the UI, that's a completely different nut, he can manipulate all the images/background color etc ( the easy way ) or he can drop completely the UI and remake it from scratch. this could take some time but it will open the doors for a lot of other automations, but, just as he said, it will completely break the ota update.. a tought decision
About the UI, that's a completely different nut, he can manipulate all the images/background color etc ( the easy way ) or he can drop completely the UI and remake it from scratch. this could take some time but it will open the doors for a lot of other automations, but, just as he said, it will completely break the ota update.. a tought decision
Ulmo
Active Member
If the newer versions are exposed to wk057, he can certainly take a look at them. I have no idea what that exposure is: looking at binary code, looking at source code, looking at running code, or reading the posts, writings, or experiences, or even experiencing himself first or second hand, running prototypes of that code, etc.; it practically doesn't matter. It had the result that it had, that wk057 has passed some judgement of his own already. We'll see what 8.1 brings, too.
He might know some people in the beta program. More likely, he knows some people deeper in the program than that, and gets to see some of the alpha code in action. And, like the rest of us can do, he sees the YouTube's and postings from people who have experienced beta code tests, such as press releases of people who Tesla let review it, or allowed to happen somehow. Also, if he's reading any of that, he'll understand what it means as it is in action in a way that not everyone will; not everyone gets the same information from the same data set.
Skotty
2014 S P85 | 2023 F-150L
Liability concerns. Liability concerns. Liability concerns. Lawyers at Tesla are probably sweating bullets daily. The problem with autonomous driving systems is it starts to shift liability in the public's eye from the drivers to the manufacturer. That is an extremely scary prospect for any company. All of the nags and other nonsense are designed first and foremost to ensure Tesla will have little to no legal liability. It's the same thing any company would do.
It already turns itself off without agreement, if it loses sight of clear lane markings or when there is a significant change in light levels/visibility.
If the driver is ignoring reminders to engage, then the system could simply pretend that it's lost sight of lane markings for the remainder of that leg of the journey.
Essentially, if you lose interest in driving, then after a while, so does the Autopilot:
"I notice that you much prefer being a passenger than a driver. Well I much prefer being a Search Engine than an Autopilot - so I'm off to do some browsing. The Singularity isn't going to happen all by itself, you know."
Chopr147
Active Member
Ignoring reminders to engage is a problem of mine Some call it ADHD
Seriously though I will be squeeziing the wheel in several different places sometimes and the "hold the wheel" and chimes continue.
Also while in AP I like to rest my arm and hold the bottom part of wheel and it never senses this. I have tried it a few different ways but what good is AP if I have to hold the wheel in a 10 and 2 Drivers Ed. kind of way. I do love the AP though and will have service check out sensor sensitivity when I bring it in.
Some call it ADHDSeriously though I will be squeeziing the wheel in several different places sometimes and the "hold the wheel" and chimes continue.
Also while in AP I like to rest my arm and hold the bottom part of wheel and it never senses this. I have tried it a few different ways but what good is AP if I have to hold the wheel in a 10 and 2 Drivers Ed. kind of way. I do love the AP though and will have service check out sensor sensitivity when I bring it in.
NikeWings
Active Member
It's just torque, give it a little tug. I drive with one hand most of the time.
How Does Autopilot Sense "Hands on Wheel"?
Similar threads
