Rolf Mantel <
news@hartig-mantel.de> wrote:
Am 05.04.2025 um 19:35 schrieb bp@www.zefox.net:
Frank Krygowski <frkrygow@sbcglobal.net> wrote:
Here's a video on Automatic Emergency Braking technology for cars, to
protect bicyclists and pedestrians.
https://www.youtube.com/watch?v=fWZQGcMN3Vc
I have no experience with that version, but I've experienced AEB driving >>> the 2022 Kia EV. It's always been responding to a false alarm. For
example, once on a 35 mph city street, the street had an oddball sudden
jog left and right. I think that put a parked car in the straight ahead
view of the camera system, and the brakes went on. At other times, I
could see and time a car slowing to turn into a driveway, but the camera >>> seemed to assume he was stopping in the road.
Anyway, if this technology becomes common, ISTM it might help reduce the >>> ~1000 bicyclist and ~7000 pedestrian fatalities in a typical year.
Seems to me that AEB system was designed for high/freeway use only.
A slow city street is a ludicrously complex environment for AEB at
the present state of the art.
I think you confuse "AEB" (Automatic Emergency Braking) with Adaptive
Cruise Control; Seems you did not watch the linked video. On
highway/freeway you rarely have pedestrians cross the road.
Not at all. I worked on automated cars in the late '90's.
The video is merely a promotional work of art advertising
"autonomous emergency braking", apparently reliant on
machine vision.
AEB currently has several false positives (plus as a "false negative" it cannot cope with the zig-zag bike route design of <https://en.wikipedia.org/wiki/Protected_intersection> ). As a pure
safety feature, you normally cannot switch it off as a user.
But if you have possibility to compare AEB systems designed in 2018 with those from 2020 and those from 2025 you see massive improvements.
Improvements, yes. Ready for general use? Not yet. Use of the term
AEB is is overoptimistic at best and dishonest at worst. Machine
vision alone still isn't close to good enough for public use. It
suffers too, even at best, from the limits of fog and darkness.
A more realistic idea would be to put cellphone recievers in cars
and program the control system to not hit the cellphone. That would
let people's cellphones function as beacons to be avoided. It'd work
for pedestrians too, and at night or in fog, provided the cellphone
is on and transmitting fast enough, which isn't guaranteed AFAIK.
One of the lessons to emerge from the project I participated in was
the value of cooperative communications between road users. Making
our cars behave sanely required them to communicate both status
(speed, for instance) and intention (desired speed) to give what
was than called "preview" of what was _going_ to happen. That
provided enough advance warning to let the control system keep up
with events. That's why we have brake and turn signals.
The goal is good, the hype is very, very bad.
bob prohaska
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)