Waiting minutes for a light that never changes
larry_scholnick
street gets a green light forever, until either traffic on the side
street is detected by the sensor or the pedestrian button is pressed.
This morning while walking I came upon the scene of a woman who had
stopped at a red light on the side street, but had driven through the
crosswalk and was stopped beyond the crosswalk. The traffic sensors
are located in all traffic lanes in the 20 feet preceding the
crosswalk and halfway through the crosswalk. She had stopped so far
forward that the traffic signal was completely unaware of her
presence.
While I was watching, a southbound car on the primary street tripped
the left turn lane sensors; the traffic signal responded
appropriately. Northbound traffic cycled to red and the southbound
left arrow cycled to green. Much to the women's dismay, the traffic
signal returned to its default state, green for north/south, and
waiting for sensor input from east/west or from the left turn lane.
Luckly for her, another car pulled in behind her, stopping legally
(before the crosswalk) and tripping the sensor.
I considered walking out to explain the situation to her, but I'm not
sure that I could have avoided being insulting, so I quietly let her
experience the consequences of her actions.
For those of you who wonder why the traffic sensors did not trip when
she drove over them (which she certainly must have done), let me
explain. Many traffic signals in Los Angeles operate on a fixed
timing sequence. Here's a typical 60-second cycle:
00 - Check sensors; if car present or pedestrian button has been
pressed since last sensor check, begin blinking red DontWalk for side
street.
10 - If cycling, solid red DontWalk, and display Yellow light for
primary street.
12- If cycling, display Red light for primary street and Green light
for minor street.
The light stays green for the minor street for 20 seconds if the
pedestrian button was pressed, or until 2 seconds of sensor inactivity
on the minor street.
32 (or earlier) - Display Yellow light for minor street.
34 (or earlier) - Display Red light for minor street and Green light
for primary street.
If you are not on a sensor at 00, the minor street cycle is not
initiated. If you are turning right, pass the sensor at some other
time in the cycle, and proceed with your right turn, the minor street
cycle is not initiated. If you are going straight or turning left and
you run the light, the minor street cycle is not initiated. Only if
you are on a sensor at the moment it is checked does the minor street
cycle get initiated.
Dave C.
Street sensors suck. I've stopped my 18-wheelers right where the
sensors should be, and many of them can't sense a 40-ton vehicle
sitting right on them. I know at least one sensor where the "side"
street (the main highway through the area, but anyway) has two lanes
going straight, and if you stop in the RIGHT lane where the sensor
should be, you will be sitting there until someone pulls up beside you
in the LEFT lane. The sensor only activates if you stop in the left
lane for some odd reason. To encourage LLB, I guess. -Dave
richard
That cycling method sucks.
At the least, it should check the sensor every 5 - 10 seconds.
Continue the light cycle normally.
If the sensor is tripped, change the lights accordingly.
In many locations, the sensors are practically checked constantly.
So as soon as a vehicle crosses one, the lights begin to change.
When I first began riding my bike in 1978, I found out quickly that many of
the sensors could not be tripped. In many cases I had to position the bike
over the corners to get it to trip.
Steve Sobol
scotte...@yahoo.com says...
>
> Last time on rec.autos.driving, "Dave C." <no...@nohow.never> said:
>
> >Street sensors suck.
>
> No, street sensors are good things. It's the implementations that
> suck.
Mojave Drive at La Paz Drive, just east of the I-15 interchange in
Victorville. Turning left from northbound La Paz? Good luck getting an
arrow, it's seemingly quite random.
Most, if not all, of Victorville's other intersections work great.
--
Steve Sobol, Victorville, California, USA
sjs...@JustThe.net
Brent
> I considered walking out to explain the situation to her, but I'm not
> sure that I could have avoided being insulting, so I quietly let her
> experience the consequences of her actions.
I try to explain this to people when I am bicycling. I stop at the light
and stop on the most forward sensor and try to trip it. A driver will
come along and then stop before the sensor furthest from the line. I
then have to try and encourage them to move forward.
I have on occasion come to a red light where someone is already waiting
but stopped so far back as to not be on any sensor. If safe, I pass them
and then stop at the line and trip the sensor. Occasionaly that driver
sitting off the sensors gets angry. Explaining their stupidity to them
is pointless.
jgar the jorrible
> Last time on rec.autos.driving, richard <mem...@newsguy.com> said:
>
> >When I first began riding my bike in 1978, I found out quickly that many of
> >the sensors could not be tripped.
>
I've posted elsewhere about multiple experiences like this I've had on
motorcycles going back to at least 1978. That was Goleta, where you
couldn't make right turns instead to get around it, as there was an
ocean in the way. In 1982, a coworker and I were both on small motos
in the left turn lane in front of the Torrance Police department,
jumping up and down on our bikes to try to get the sensor to work.
Eventually, we just made a U turn, and sho' nuff, got pulled over.
They separated us and questioned us, then finally let us go when we
consistently told the same story. The next morning on the way to
work, the other guy got nailed by a red light runner who had been up
all night on PCP. I always wondered whether the police made some
untrue connection about that.
Any implementation must account for bad sensors.
jg
--
@home.com is bogus.
Lexus sudden acceleration lawsuit:
http://www.autoevolution.com/news/mccunewright-files-national-class-action-against-toyota-12980.html
Patrick Scheible
> Last time on rec.autos.driving, richard <mem...@newsguy.com> said:
>
> >When I first began riding my bike in 1978, I found out quickly that many of
> >the sensors could not be tripped.
>
> They had loop sensors way back in 1978?
Yes, there were some near my house in Palo Alto by then, though they
were fairly new.
-- Patrick
Josh
<scotte...@yahoo.com> wrote:
>I'm surprised she didn't just back up, if not to trip the sensor then
>to at least get behind the limit line so some passing cop didn't give
>her a ticket.
Sounds like she had no clue that the way she was driving was wrong,
whether to trip the light sensor or to be legal in the first place.
And probably wouldn't have even if it was explained to her.
>So bottom line you're saying that the sensors are POLLED, not
>INTERRUPT-DRIVEN?
>
>Any first-year Computer Science student knows how LAME that kind of
>design is. Then again, that doesn't surprise me.
>
>What SHOULD happen is this: whenever a vehicle drives over a sensor,
>an exception is generated at the microcontroller and the event is
>counted. Then, at t=0 (or whenever), the controller checks the event
>counter (and then clears it to 0 for the next cycle). BINGO! Problem
>solved!!
I'm pretty sure this designed this way on purpose to minimize
interruptions to the main traffic flow (isn't it annoying to be in a
pile of cars approaching from the last light and have a light change
*right then* to let *1* car go, when waiting 20 seconds would have
minimized everyone's time?). Specifically, the microcontroller's
logic is saying "I saw a car go by, but it's not there anymore, so it
must have turned right on red. No need to change"
I'm quite certain the standard controller can also start the change
instantly if programmed that way, as many around here (Sacramento) do
that.
The algorithm can be set up however the traffic engineers think best,
based on the inputs from the at-intersection sensors, approaching
mid-block sensors, cycle timers, time of day, etc.
>
>>If you are not on a sensor at 00, the minor street cycle is not
>>initiated.
>
>What's REALLY asinine is that, if you are in a left-turn pocket, you
>can be present at the sensor at t=0 and STILL not get a green arrow
>for that cycle. That's because the sensor is only checked at t-15
>seconds, or 15 seconds BEFORE the start of the cycle.
I agree that's annoying. I find that just encourages me to race up to
a light at full speed, braking at the last second, in hopes of getting
the light a cycle sooner.
>It's clear that some pretty moronic design decisions went into the
>sensor-controlled traffic signals in SoCal. It's almost as if the
>designers are anti-car; they ceretainly don't seem to have the
>"first-come, first-served" sense of fairness in mind when they program
>these things.
Sensors in general are more car-friendly than fixed-timing lights
(except where fixed timing allows a non-stop progression block to
block on city streets), but I agree that sometimes the algorithms and
timings chosen are anti-car. In Folsom, there's an intersection with
a *very* short straight-ahead light, regardless of the number of cars
waiting, because they don't want people cutting through a
neighborhood. The secret is that turning right at that light then
left 1/8 mile later gets you on a parallel street (which is even more
residential) that joins back at a stop sign 1/4 mile ahead. Usually
well before the people who got shafted out of the first cycle get
there.
Josh
gpsman
> Last time on rec.autos.driving, Josh <no_need_to_s...@nobody.org>
> said:
>
> >I'm pretty sure this designed this way on purpose to minimize
> >interruptions to the main traffic flow (isn't it annoying to be in a
> >pile of cars approaching from the last light and have a light change
> >*right then* to let *1* car go, when waiting 20 seconds would have
> >minimized everyone's time?).
>
Don't you have a shorter list?
> There was a time when traffic signals were controlled solely by timers
> and the timers were set such that a car moving at the posted speed
> limit could hit a reasonable number of green lights all in a row. Now,
> with sensor-controlled lights, it's virtually guaranteed that every
> intersection will have a RED light by the time you get to it, even if
> the intersection you just left has just turned green.
Where will you take your false premise this time...?
> There will
> almost always be that 1 car on the cross street to trip the signal red
> for you.
So, what's the rush?
> There does seem to be a timer that measures the amount of time since
> the last car tripped the sensor; if another car trips the sensor
> before this timer expires, the light remains green. If the timer
> expires, the cycle ends at that point.
That could not be the result of a default green interval and polling
of the sensors?
> What this means is if you're a
> "loner" on the road and drive with a large space cushion around you,
> you're going to get a lot of red lights, whereas if you tailgate or
> drive in a big clump of cars, you'll get more greens because you'll
> arrive at the sensor only a few seconds after the previous car tripped
> it and you'll keep the "window" open.
"it's virtually guaranteed that every intersection will have a RED
light by the time you get to it, even if the intersection you just
left has just turned green."
> >The algorithm can be set up however the traffic engineers think best,
> >based on the inputs from the at-intersection sensors, approaching
> >mid-block sensors, cycle timers, time of day, etc.
>
> Unfortunately, "fairness" and "first-come, first-served" don't appear
> to be on the list of "best practices" for traffic engineers. :(
Those stupid fucks are probably more focused on moving all traffic
most efficiently.
> >I agree that's annoying. I find that just encourages me to race up to
> >a light at full speed, braking at the last second, in hopes of getting
> >the light a cycle sooner.
That sounds like it took a lot of cost/risk/benefit analysis.
> Same here.
No shit?
> That's also why being a "Sloth Coaster" is a losing
> strategy: the later you arrive at the intersection, the less chance
> you have of triggering the sensor within its "window of opportunity"
> and the greater the chance you'll have to wait through a needless
> extra cycle.
That's only true if there is no traffic to the front to trip the
sensor for you, depends on what one considers "losing", and your
premise requires the "sloth coaster" to misjudge the "window of
opportunity".
("Sloth coaster" is your personal term which you have yet to define
and is therefore meaningless, but it fits perfectly within the vast
majority of what you write.)
Besides, "it's virtually guaranteed that every intersection will have
a RED light by the time you get to it, even if the intersection you
just left has just turned green."
You refute your own arguments, in the same post.
Again.
-----
- gpsman
Paul D. DeRocco
>
<snip>
>
> For those of you who wonder why the traffic sensors did not trip when
> she drove over them (which she certainly must have done), let me
> explain. Many traffic signals in Los Angeles operate on a fixed
> timing sequence. Here's a typical 60-second cycle:
>
> 00 - Check sensors; if car present or pedestrian button has been
> pressed since last sensor check, begin blinking red DontWalk for side
> street.
> 10 - If cycling, solid red DontWalk, and display Yellow light for
> primary street.
> 12- If cycling, display Red light for primary street and Green light
> for minor street.
> The light stays green for the minor street for 20 seconds if the
> pedestrian button was pressed, or until 2 seconds of sensor inactivity
> on the minor street.
> 32 (or earlier) - Display Yellow light for minor street.
> 34 (or earlier) - Display Red light for minor street and Green light
> for primary street.
>
> If you are not on a sensor at 00, the minor street cycle is not
> initiated. If you are turning right, pass the sensor at some other
> time in the cycle, and proceed with your right turn, the minor street
> cycle is not initiated. If you are going straight or turning left and
> you run the light, the minor street cycle is not initiated. Only if
> you are on a sensor at the moment it is checked does the minor street
> cycle get initiated.
How do you know that's how the algorithm works? Sounds like a reasonable
deduction, but it's not like a more robust algorithm would take more
computing resources, so it's hard to believe that an expensive system like
that would be designed so badly.
--
Ciao, Paul D. DeRocco
Paul mailto:pder...@ix.netcom.com
jgar the jorrible
> > "larry_scholnick" <larry_scholn...@yahoo.com> wrote
FWIW, http://www.mccain-inc.com/traffic/item/local-control-software/program-2033.html
They're in Vista, CA. I was surprised to see the reference to my city
in wikipedia traffic signal article.
jg
--
@home.com is bogus.
http://www.signonsandiego.com/news/2009/nov/13/driver-pleads-not-guilty-in-fatal-vista-accident/
Paul Hovnanian P.E.
stop for a red light at a sensor-triggered intersection, you could sit
there all day. You've got to roll your vehicle across the crosswalk and
practically into cross traffic to get them to work.
*Where they've cut existing pavement to install them, you can see the cuts.
--
Paul Hovnanian pa...@hovnanian.com
----------------------------------------------------------------------
Have gnu, will travel.