Tend to agree but to say friction braking vs regen braking is fail is a Bit of a stretch in my view. My car (using any means to stop) comes to a smooth consistent stop as it should. Which is at base level what it needs to do and I am fine with that. Not saying he is Wrong in his opinion just saying at this layer of software we simply want it to drive and perform safely. The Limo max efficiency version can take years which I can except.no such thing as a "qualified disengagement".
When someone chooses to disengage FSD is up to them. If the assistance is no longer assisting then it needs to be disengaged.
Disengaging to prevent late braking is a personal choice which in no way invalidates a disengagement just because it doesn't meet somebody else's definition of what "counts" as a disengagement.
95% of my drives with multiple disengagements would be a zero disengagement drive from Omar
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
FSD Beta 10.69
- Thread starter Buckminster
- Start date
AlanSubie4Life
Efficiency Obsessed Member
Does it do this consistently? Evidence required for this claim. Please provide several non-selective examples on a slight downhill grade (where it struggles the most as discussed - but failures occur on level grades as well). Uphill grades usually are successful though not minimum jerk.
I think you want the software to drive like a good human to minimize the chance of rear end collision. Stopping more gradually and smoothly reduces the chances of that type of collision.
This is not only a comfort issue, it is also a safety issue.
I need to try out some other Tesla's out there, because clearly mine isn't behaving the way some are claiming it does for them. My threshold for what's comfortable and uncomfortable is just different I guess, and I don't expect the system to drive exactly like me. Also, you like to compare it to a "good" human, but humans all have vastly different driving styles with lots of variability. With a cluster of 20 cars all heading the same direction towards a red light, you're going to get different deceleration curves - they're not all going to get to the light at the same time with the same curves. There's the guy who slows down really early (I call them "sneaking up on the red light"). There are guys who race up to the red light and slow down aggressively, because they want to get to the light first for some reason (I call them "win at all costs"). And everything in between. My car doesn't race up and slam on the brakes, and it doesn't creep up to the light either. On a bell curve, it's right in the middle section, which I'm fine with.
I guess a good test would be a cup of coffee - does the car slow down for an upcoming red light hard enough to spill a cup of coffee?
AlanSubie4Life
Efficiency Obsessed Member
Can probably do this test with water. Will put on list.
Jerk and peak acceleration, deceleration specifically (more relevant here but correlated with high jerk) can actually be quite high. Average deceleration is only slightly higher (absolute value to avoid confusion in terminology and sign) than mine probably, and in some cases lower while still extensively using the friction brakes (due to large amount of lost regen time as discussed).
Use a water bottle - filled to a point, then draw a line above that point with a sharpie. It would have to be a reasonable distance to the line - say 1 inch? What level would an average person feel comfortable driving with a cup of liquid with no top on it? It would have to be enough so that if you brake hard, the water easily crosses that line.
AlanSubie4Life
Efficiency Obsessed Member
We could also use the regen power bar and account for speed but I guess that is too easy???
I guess some people don’t trust the display and want something more tactile (and the water bottle will remove velocity dependence of course).
Really just need a g-meter with readout on all axes over time.
Last edited:
The methodical process you layout is indeed the traditional approach, but Elon doesn't think in traditional approaches. Can his entrepeneurial approach of "fall -> fall -> succeed" work in a scaled company? Time will tell, but so far his approach has "wham-bammed" Tesla to a $300B+ market cap company, even when the "fails" burn a few customers along the way.
Your regen bar and my regen bar might not be the same - a Tesla with a smaller battery pack may not take the same amount of energy at the same speed - or some cars have different weight which affects how fast they'll slow down. We'd have to establish a base-line first - such as doing a test to see how fast your car slows down vs my car on pure regen.
Example:
On a flat, straight surface, going 50MPH, we take our feet off the accelerator at a specific point and measure how far we travel before coming to a complete stop. This is assuming a pre-conditioned battery at 50% SOC on a 70 degree day to ensure there is no reduction in regen.
I'd be curious if the results would be identical, or if there would be variability in stopping power that's outside the margin of error.
PACEMD
Active Member
What if it was coffee from McDonald's rather than water, and you could sue if it crossed the line? I'd be ok with that, as long as I won enough money to hire a chauffeur.........
Last edited:
it must be environmental. I've read enough folks saying they don't experience these issues. But traffic light stops in my area are rarely smooth and much more likely to have my passengers pressing the imaginary brake pedal in front of them. Circumstances where this happens are when there are other cars in the same lane, approaching lights at >30mph, or approaching lights after corners. Lights on 65mph roads are the worst, it either starts braking for warning sign then speeds up again, or just plain brakes really hard. The only time it gets better is when the speed limit starts reducing before the light, but they are few and far between here.
PACEMD
Active Member
Hello! McFly! Why do you think Musk is pretending to lose everything through Twitter? Right, exactly, lightbulb finally goes off.........so when he escapes to Mars with all the world's money, every last penny, no one will suspect him or his FreeIlluminatiMason handlers........duh......
AlanSubie4Life
Efficiency Obsessed Member
This is important. It is much less of an issue at lower speeds, even on downhill grades.
This 69 is not as deserving of the moniker as hoped.
Knightshade
Well-Known Member
There is, of course, no actual NHTSA regulation that requires driver monitoring at all
So it's a weird thing to be reporting as being "investigated"
"Hey we investigated and found out the thing you want to remove isn't legally required anyway so I guess we just wasted everyone's time...sorry bout that!"
Then again it's BI, they have a weekly quota on tesla hit pieces they have to meet, and is a publication founded, and still to this day run, by a dude who was permantantly banned from the securities industry (in addition to his fines) for the SEC fraud charges against him.
Last edited:
Frankly I could care less what the regen power bar shows. For now if the stop was reasonably smooth I'm good. Should Tesla maximize regen of course but in no way do I mark down a drive because of that. In a year probably.
I like this; it inspired me to create an image. My first crack at Stable Diffusion:
Crew Piggy StarBank on Mars:
(a collaborative effort of SpaceX and the Boaring Company)
Daniel in SD
(supervised)
I’m a little confused by this discussion. Is FSD beta using friction brakes instead of regen or is it decelerating faster than regen is capable of?
Driving manually without using friction brakes is easy but even regular Autopilot seems to struggle with it.
Driving manually without using friction brakes is easy but even regular Autopilot seems to struggle with it.
AlanSubie4Life
Efficiency Obsessed Member
It’s a good question. Based on videos, I believe it is decelerating faster than regen is capable of doing by itself (meaning g forces are greater than they ever would be at a given speed (and slope), under manual control). However, I’d have to study bars of similar stops under manual control (no input) very closely to figure it out.
In addition there is the possibility that FSD can use more regen at a given speed than normal driving, in any case. No idea whether that is true - I think it is of course likely that it can eliminate the “feathering in” that a release of the accelerator will initiate when under manual control. So two slightly different things there.
When I post the video without intervening I guess you can study it.
My belief/guess is it is just requiring too much deceleration and thus using brakes.
When I see that braking component on the screen, it never “feels” good. So that is why I believe that.
Daniel in SD
(supervised)
I should loan you my VBOX so we can get to the bottom of this. We need jerk plots!
I don't know if this matters at all, but with the weather being colder, my regen bar has several ... in the morning. But when I enable Driver Assist or Advanced Driver Assist, those ... disappear. It must be using friction brakes, since regen is quite limited with 6 or 7 dots.
Similar threads
- Article
