If it could be adjusted somehow on the spot for extreme left and right lane driving and for trucks (momentarily) it would be a safer feature but that is only relevant in modest to heavy traffic. It is perfect only for center lane driving. And is ok in all lanes in light traffic.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Things my wife said about Navigate on Autopilot tonight
- Thread starter AlanSubie4Life
- Start date
I'm sure I'm just imagining this, but whenever I pass a semi, my car seems to get closer and closer to the right side of the lane. I really wish Tesla would program in a bit of opposite lane hug when passing trucks... or even just passing other cars in general.
Hope they didn’t teach you that, though it would explain the behavior I see in places.
That is not nearly enough distance. A car length is 15 feet. At 70mph that would be 105 feet. At 70mph you are traveling 103 ft per second.
So that is only a 1-second following distance/time.
So, you need more like 3 car lengths per 10mph.
Really these rules of thumb are slightly flawed if you think about the underlying physics (stopping distance goes up with the square of speed, not linearly...stopping time is a linear function of velocity...both have an offset for reaction time though) but 3 seconds is pretty safe at most reasonable speeds if you are paying attention, unless the person ahead runs into a brick wall. At much lower speeds you can probably safely follow a bit closer than 3 seconds, due to the physics.
Last edited:
DopeGhoti
Active Member
The rule of thumb I was taught was a four-second following distance. Though in practice, I rarely live up to this; I'm usually at around a 2-3 second following distance which, every time I notice this, I momentarily realize may be much to my chagrin someday.
My wife rarely compains about my driving, although she does frequently complain about the temperature being too cold in the cabin - she likes it at 78 degrees or higher. At this temperature, she generally falls asleep within 15 minutes of being on the road. I've taken sharp curves at over 2x the posted speed, and she hasn't said a word. Generally, AP's jerkiness does not seem to wake her up. I think I'd rather have a companion who is awake and alert during trips - especially at night after long outings. On the other end of this spectrum is my mother who functions as an early alert system that makes ME uncomfortable. Wouldn't mind a slider on AutoPilot from "responsive" to "asleep at the wheel"
I am not challenging you on that but hearing it on the forum and reading it in the Tesla documentation are two completely different things. The 1 car length for every 10 mph is avg for cars, as motorcycles can stop infinitely faster than a car theirs would be radically different, as would the following distance for an 18 wheeler. I do not trust an arbitrary number like 1-7. A Tesla semi and truck and X and 3 would all have much different stopping distances based on their gross weight. So, minimally, Tesla needs to elaborate on what the 1-7 stopping distance is indicative of.
I believe you are confusing this with the distance to stop for a brick wall. Clearly the rule-of-thumb is an avg based on an avg car you are in and following. A motorcycle for instance could likely tailgate a semi as it can stop infinitely faster than a semi. Reverse that and a semi must lag a motorcycle way greater than 1 car length for every 10 miles/hr. This is what decades of people were taught. A car cannot stop instantaneously. The rule-of-thumb also assumed an avg response time. Also, did the car / semi / motorcycle in front tap the brakes or slam them on? ITS A RULE OF THUMB way better than pick a number between 1 and 7.Hope they didn’t teach you that, though it would explain the behavior I see in places.
That is not nearly enough distance. A car length is 15 feet. At 70mph that would be 105 feet. At 70mph you are traveling 103 ft per second.
So that is only a 1-second following distance/time.
So, you need more like 3 car lengths per 10mph.
Really these rules of thumb are slightly flawed if you think about the underlying physics (stopping distance goes up with the square of speed, not linearly...stopping time is a linear function of velocity...both have an offset for reaction time though) but 3 seconds is pretty safe at most reasonable speeds if you are paying attention, unless the person ahead runs into a brick wall. At much lower speeds you can probably safely follow a bit closer than 3 seconds, due to the physics.
No. I was not. I was just saying your statement (which is probably not what you do - you probably allow more distance) results in about a 1-second following distance at 70mph, which most drivers’ manuals would deem insufficient regardless of the vehicle capabilities.
In general I agree with your statement that using a following time is more appropriate than using a fixed distance. And that is what TACC provides - a following time.
More on topic for this thread, I was using following (time) setting 5 for the above video.
It is in the manual. And it is easy to measure the effect yourself if you doubt the manual.
Fernand
Active Member
No, I don't think the current recognition system can interpret those advisory speed limit signs that are hanging on the side of the road, like when you approach a curve. That's beyond the stage of recognizing stop signs and red lights.
I think one of the big differentiators in how people see the Tesla automation, I mean between those who are angry at shortcomings and those who are not, is expectations. And that's based on past experience.
I just had a very disenchanting experience in buying some personal electronics. We live in a time of fading ethics. Quick buck, screw the rest. There's always been those players, but it's spreading, and that makes a lousy bed of past experience. I'm old fashioned enough that, as an engineer, when I see the difficulty in implementing self-driving my first thought isn't "they cheated me". It's more like, it's harder than anyone expected, of course it's behind schedule, we're working on it (I mean we as a species), we're slowly getting there. And I look at how much of the total puzzle Tesla is handling, and I think Tesla is getting there first. I can use what they've done so far. So why sweat it?
I think one of the big differentiators in how people see the Tesla automation, I mean between those who are angry at shortcomings and those who are not, is expectations. And that's based on past experience.
I just had a very disenchanting experience in buying some personal electronics. We live in a time of fading ethics. Quick buck, screw the rest. There's always been those players, but it's spreading, and that makes a lousy bed of past experience. I'm old fashioned enough that, as an engineer, when I see the difficulty in implementing self-driving my first thought isn't "they cheated me". It's more like, it's harder than anyone expected, of course it's behind schedule, we're working on it (I mean we as a species), we're slowly getting there. And I look at how much of the total puzzle Tesla is handling, and I think Tesla is getting there first. I can use what they've done so far. So why sweat it?
Yep, I found it, thank you.
This is kind of an interesting subject. I am almost 100% positive the rule was 1 car length for every 10 mph. So I looked up what the 'woke' generation is telling teenagers.
The following distance is a space between your car and the car ahead of you. It is recommended to keep a reasonable following distance so you can safely stop in a case of an emergency, e.g., if the car ahead of you stops suddenly. A defensive driver maintains a safe following distance of at least three seconds behind the vehicle ahead and increases it depending on weather and road conditions.
Here's the problem though. First, in sentence 2 "if the car in front of you stops suddenly. That is physically impossible. They may brake suddenly but they won't stop suddenly, unless they are going sufficiently slow that the coefficient of friction is sufficient to bring them to a jerking immediate stop, ~1 mph? The last sentence is silly. How is anyone, much less a 16 yr old, going to measure what a 3 second stop distance is. Seconds are a measure of time, distance is a measure of space. However, that is similar to the language in the owners manual. The only way that makes any sense at all is if stopping was instantaneous. Put simply, so long as the following car is going equal to or slower than the followed car they will never ever make contact. That one second you speak of really equates to reaction time. Can the following driver upon seeing a brake light have his / her foot on the brake w/i one second. If the driver is, at all, paying attention that's an easy bar to clear. Once your foot is on the brake all the following driver need do is insure they are going slower than the braking car in front of them.
My point in this thread was Teslas have enough smarts to know 1) how fast they are going, 2) how far away they are from the followed car, and 3) the acceleration relative to the followed car. So long as the acceleration is less than or equal to 0, it's all good. Negative acceleration is the technical term for what is popularly referred to as deceleration.
In a future highway system where only autonomous cars are allowed, the followed car could be virtually on the bumper of the followed car because the reaction time to a non-zero acceleration would be microseconds. In that prism following distance is a courtesy to the followed driver. I suspect in the next 12 months Tesla FSD cars will manage their own braking.
As for the verbiage in the owners manual, if that's what they meant, (the time it takes to establish a negative acceleration relative to the followed car) that is not what they said.
Happens all the time, effectively. In a serious rear end collision a car stops in 15 feet or less.
Absolutely possible, see above.
The one second is not just for reaction time, I assure you. (One second is a very slow reaction time.) If you encounter an emergency you will want more than 1 second of following time.
Keep in mind your following distance is also determined by the vehicle behind you, its following distance, and its capabilities, as detailed in many drivers’ manuals.
Anyway, back on topic, regardless of the chosen following time, I think my wife wishes Autopilot could maintain that chosen time in a smooth fashion.
Speaking of which, something to try...
Drive down a road that you know the AP struggles with, as far as speed management, with your AP engaged. Now drive the same piece of road again, but only use TACC. Any difference?
Discuss.
I'll go first...
In my very limited experience, it seems like TACC handles accel/decel differently and has different phantom braking characteristics than AP does.
I drive the same stretch of road frequently (as I think a lot of guys do), and there's about a one mile stretch on this road where I get a ton of phantom braking, etc while using AP, yet TACC is butter smooth. Even just following behind cars while using TACC is more smooth than AP.
I'm sure there's going to be a lot of responses of, "nope, they're the same." For those of you with that response, maybe try to make a concerted effort to really analyze the difference going forward?
It's almost like AP uses a specific subset of TACC variables that are set differently when AP is driving vs when the person is driving.
Speaking of which, something to try...
Drive down a road that you know the AP struggles with, as far as speed management, with your AP engaged. Now drive the same piece of road again, but only use TACC. Any difference?
Discuss.
I pretty much exclusively use TACC, and over time they keep adding "capabilities" to it that I don't want.
Like it will phantom brake if it thinks the car in front of me is going to cut in. I don't want that as my entire reason for using TACC is to simply have a smooth throttle control and to stay at a set speed. I can manage the watching out for other traffic bit.
I really want TACC to be very simple, and that's how it used to be.
Heck in the very beginning with AP1 the TACC system didn't even slow for corners. In those it was fairly smooth after the first few updates.
Sure there were some complaints time to time, but I think that was mostly regarding personal preferences versus downright unsmoothness.
With AP2/AP2.5 there have been times with fairly smooth TACC, but then some update ruins it.
It's really quite sad because with an EV the adaptive cruise control system should be exceptionally smooth. There is no reason for it not to be.
The reasons it's not is Tesla is prioritizing other things like trying to brake for stopped objects, braking for cut in's, etc. All that crap that I don't want it to do as those things are my responsibility since I picked TACC.
I don't mind it trying to do a lot more when someone uses AP or NoA, but I really wish they would have left TACC alone.
Now is the behavior exactly the same? No. NoA will suddenly try to slow down if it thinks you need to get over even if you've declined the lane change 100 times. So NoA has the potential of being much worse.
Are you referring to coming to a stop because there was an instantaneous change in mass like the following car rammed into a stopped vehicle and there were only 15' of skid? A car slamming into a building yeah, might come to a rest in 15'.
I'm sufficiently removed from High School and College Physics to 'do the math' from memory but stopping (stop being 0 velocity) in 15' they'd have to be going incredibly slow (or have an instantaneous change in mass). Even if the brakes locked up the momentum of the mass of the car would carry it. Yes, the friction at that point would be high but so is the mass of the car.
From How to Calculate Skid Speed
Calculate the skid speed by multiplying the drag factor times the braking efficiency times the skid distance times 30 and taking the square root of the result. For example, if the car skidded 120.5 feet on a road with a drag factor of 0.5 and left four skid marks, the car was going about 42.51 mph.
sqrt(.5 x 15 x 30) = 15mph in your example. NOT a serious accident. Cars do not stop instantaneously, unless they are going incredibly slow or hit immovable objects.
I am not trying to troll. I think this is an interesting topic of conversation and it involves physics and math.
Within a class (loaded 18 wheeler, passenger car, motorcycle) all members of that class stop +/- in about the same distance given the same initial speed. A motorcycle could almost safely 'draft' behind an 18 wheeler because the stopping distances of each are so radically different. The reverse is monumentally not true for the same reason.
I think the Tesla emergency braking facility is an outstanding feature except the degree of braking should be commensurate with the acceleration relative to the leading object, vehicle or building/fence. The Tesla just needs to slow down in order to make closing velocity 0. This would work for moving vehicles and stationary objects, fences and structures.
Good conversation though!
As not everybody on here has a physics background
F = MA is your friend.
acceleration = change in velocity or direction.
Yeah, I totally agree!
Like “buy FSD before August 16th, 2019, when we will increase the price!”.
“Your car will pick you up anywhere in the parking lot, really. For the low, low price of $6000. Buy it now!!!!!!”
Yes. For example, a Miata you are following at 1 second delay piles into the back of a stopped UPS truck.
x = v^2/(2*a) + t_r*v
t_s = v/a + t_r
a = ~9.5m/s^2 for Model 3 with MXM4s
t_r is reaction time, t_s is stopping time
x is stopping distance
Similar threads
