01-20-2023, 02:17 AM | #111 |
Lieutenant Colonel
805
Rep 1,790
Posts |
The several times that it has happened to me it's a very abrupt and out of thin air experience and is very short lived. It slams on the brakes out of no where and immediately recovers within less than a second. What's the most concerning is that if someone were tailgating me, which is a common occurrence where I live, they would most likely rear end me based on how hard it hits the brakes.
The issue is that even though I have both hands on the wheel and am paying attention it happens quicker than any human can recover from by steering or by hitting the gas pedal. Less than a second from slamming on the brakes to recovering from the false detection. By the time you realize it happened it's already released the brake. It's one of those "What the FU#K just happened?" moments. It sounds to me like others are seeing a much longer false braking period where there is time to override it. Not so in my case. Additionally, this safety feature is unacceptably slow to react to changes in the traffic around the vehicle. Both my BMW and my wife's 2022 Mazda CX5 react in the same delayed fashion. It sees a car ahead that's getting ready to make a turn and throws up the red car in the display and HUD as you get close and continues braking long after the car has already made it's turn. Both of these vehicles have the reaction/response time of a 150 year old tortoise. I NEVER trust or rely on any of this accident prevention or autonomous driving technology. |
Appreciate
1
Blue87375.50 |
01-20-2023, 10:50 AM | #112 | |
Private
59
Rep 92
Posts |
Quote:
What typically happens is a combo of real world driving and simulated driving, along with generated images based on both to prevent that one pixel shifts from having a large impact. It's not practical to cover it all, however, because there's too many inputs to simulate (death of the universe before it's calculated levels). Then you have a system with uncertainty estimates and now you need a decision making logic to help. Fuzzy logic can work in some circumstances, but fuzzy rules become hard to tune. Suppose you find an edge case that failed. Now you have to go modify the rules, but that has a side effect on other cases. A better system in theory is a Markov decision process where there is isolation when decisions for single states are updated based on new information. The problem there becomes that the space is not well partitioned or easily tested because subtle changes in state can cause large differences in control outputs. The bottom line is state estimation and decision making techniques are very old and well studied, but no one has come up with a way to prove that the system will work reliably in all circumstances. |
|
Appreciate
0
|
01-20-2023, 01:45 PM | #113 | ||
First Lieutenant
185
Rep 355
Posts |
Quote:
Without getting into the weeds too much, there are many ways to do fuzzy logic on large input spaces without necessarily having a gigantic cross-product calculation space. As one example, that's mainly what deep learning networks are designed to achieve (ie: arrive at a pseudo-partitioned input space model which accurately captures the differentiating features of the input *******. There is also not generally a binary output from the models themselves (ie: the model doesn't say there is/isn't an obstacle, but rather there's a X% chance of an obstacle). I'd agree that lots of testing is needed; I'm opining that I would prefer a system which was tuned to do nothing in the case of uncertainty, rather than the current model which seemingly reacts if it determines there is possibly an obstacle. I find that for my own driving, the system as currently tuned, which I cannot disable, makes my own usage of the vehicle less safe, and I think that's something which should (and ideally will) lead to some large liability settlements, and ultimately a course-correction in the system design. Edit: I have no idea why the forum censors "space", but maybe it'll be visible in this edit?
__________________
Quote:
|
||
Appreciate
1
jBwFellow37.00 |
01-27-2023, 02:40 AM | #114 | |
New Member
0
Rep 6
Posts |
Quote:
|
|
Appreciate
0
|
02-23-2023, 05:22 AM | #115 |
New Member
0
Rep 6
Posts |
|
Appreciate
0
|
05-17-2023, 04:07 PM | #116 |
Private First Class
52
Rep 108
Posts |
I had one of those "out of thin air" EMERGENCY STOP issue today with my new 2024 X5 50e. I was not using any of the Driving Assist technology. Just driving down a 2 lane road with no other cars within 1/4 mile of me. I've been trying to read everything I can find which brought me to this thread. My question is: Does the on board computer indicate WHY the Emergency Stop sequence was initiated? I would think somewhere in their software it would indicated what triggered the event.
|
Appreciate
0
|
07-07-2023, 09:18 AM | #117 |
Private
66
Rep 73
Posts
Drives: 2022 BMW X5 xDrive40i
Join Date: Jan 2022
Location: Texas
|
That must have been terrifying.
I have maintained a healthy level of awareness and suspicion even over the Drive Assist features in my X5. While I do enjoy the adaptive cruise control with distance control, lane keep etc and the view on the digital gauge etc. I have turned off all steering intervention features. The sudden and 99% mostly unnecessary corrections it has tried on me have not been subtle; sometimes pulling me into the next lane without need so I turned it off. I have not experienced a full or near auto-full stop or a dramatic speed reduction such as the one you’ve described but reading this confirms my suspicions about what could happen. I was not an “auto-pilot” advocate before and definitely am not now. I highly recommend you stay engaged throughout your driving experience. I recommend making an appointment and spending time talking to BMW Service, their Genius’ and have them run a complete inspection of your systems… think you will be okay but just to have some clarity and peace of mind going forward |
Appreciate
0
|
07-21-2023, 11:09 AM | #118 | |
Major
315
Rep 1,448
Posts |
Quote:
The only new thing of late is that new-to-me Chattanooga, TN BMW did a firmware update a few days ago to 7/2023.23. It is also the case that my a/c is today not working properly even after I confirmed the correct settings. In any event, if the a/c remains a problem I will have to give CHA BMW another try at getting things right.
__________________
2019 BMW X5 xDrive 50i M Sport: ZD8/ZLS/ZMP/ZP2/ZPK/ZPX/337/ZX7/2NH
2015 BMW X5 xDrive 50i/XL/CW/DA+/EP/LP/FVS retired; 2008 X5 4.8i; 2012 X5 5.0i retired, 2004 X5 4.4i retired R360c + EL, V1G2 + V1Driver, Uniden R8 |
|
Appreciate
0
|
07-27-2023, 08:35 AM | #119 | |
New Member
3
Rep 10
Posts |
Quote:
|
|
Appreciate
0
|
07-27-2023, 09:07 AM | #120 |
Lieutenant Colonel
981
Rep 1,560
Posts |
I have not had this happen to me on the highway, but it is my understanding that you should be able to override the auto braking by pushing on the gas pedal. Does anyone know if this is true or not?
Edit: Just found this in the manual Brake interventionThe warning prompts the driver to intervene.
__________________
2002 M Roadster, Steel Gray Metailic, Gray Nappa Leather, HT, TC Kline DA Shocks & Camber Plates, w/H&R Springs, Stromung Exhaust, SSR Type C. Looking for new home.
2022 Z4 M40i, Misano Blue Metallic, Driver Asst Pkg, Prem Pkg 2023 X5 M50i, Phytonic Blue, Black Extended Merion Leather, Driving Asst Pro Pkg, Park Asst Pkg, Exec Pkg, Climate Comfort Pkg. Last edited by COKen; 07-27-2023 at 09:24 AM.. |
Appreciate
1
WagonFan271.50 |
07-27-2023, 09:10 AM | #121 |
General
13275
Rep 19,805
Posts |
Yes, the automatic braking stops as soon as you hit the accelerator.
__________________
2021 BMW G20 M340i xDrive - Verde Ermes/Black - 03/2024.40
2019 BMW G05 X5 xDrive40i - Phytonic Blue/Cognac - 11/2023.50 |
Appreciate
1
thereef5101165.00 |
03-07-2024, 08:21 PM | #124 |
Registered
0
Rep 1
Posts |
I have had my car for 6 weeks. Never had a BMW before, but my husband has and he talked me into getting one. This just happened to me 2 days ago going about 40 with nothing in front or behind me (thank goodness). I called BMW and the service manager said they could “check for a malfunction” but obviously this has been an issue for a while. Has anyone had a resolution for this?
|
Appreciate
0
|
Post Reply |
Bookmarks |
|
|