Is FLARM helpful?

What James may be referring to is the fact that Flarm indications are in
relation to your ground track not heading. The extreme example is if
flying in very strong winds, say in wave, and actually going backwards.
Another glider coming from your 6 o'clock will actually show as head on.
This effect is still there in any cross wind.
This is a fundamental flaw in Flarm, which surely could be much improved by
building in a wind algorithm to correct the indication nearer to heading.
All navigation programs have them.
Dave




At 03:22 27 November 2015, jfitch wrote:
On Thursday, November 26, 2015 at 2:45:07 PM UTC-8, James Metcalfe wrote:
At 21:29 25 November 2015, jfitch wrote:
I am curious about your experience with the interface. Did you just
have
the BF display, or was the Flarm also displaying targets on a
moving=20
map?

I have not had any Flarm warnings that would have saved a midair (3=20
years
with it now). But I have always had Flarm targets displayed on the
glide
computer moving map, and so nearly always see them long before they=20
could be
considered a threat. Two times in those three years I have gotten an
unexpected warning, though not close enough to require action to
avoid=
=20
collision.
I use those events as a learning experience to see how I need to
change=
=20
my
scans and operations, so that they do not happen again.

From this I conclude that the situational awareness, far away from an
potential conflict, is more valuable for preventing potential
conflicts
=
in
=20
the
first place than the actual warning facility.=3D20

Countless times that Flarm has identified and displayed glider within
a=
=20
km
of me that I had not seen and might never have seen.
=20
I have used only the 'butterfly' display. To be clear, of course I
am=20
seeing target alerts all the time. That is not a problem, and even=20
occasionally interesting (such as when someone starts following).
The problem is with false collision warnings. Many pilots appear to
(and=
=20
some of my pupils certainly do) find it almost impossible to resist
turni=
ng
=20
away from the Flarm direction of the 'threat', before they have seen
the=
=20
target. That is dangerous, as I wrote in a post yesterday. And note
that=
=20
the Flarm direction is often significantly different from the true
direction=20
(occasionally diametrically opposite), as it is track-based, not
heading-
based.
=20
I can see that my experience of Flarm in a very busy environment
(the=20
French Alps) will be very different from that of those flying largely
in=
=20
isolation (such as flatlands, particularly with what I would regard as
hi=
gh
=20
cloudbases (I'm a Brit!))
J.

James, if you are plagued by false alarms coming even from the wrong
headin=
g, I am all the more curious. I have not had any false alarms, perhaps
some=
false negatives (probably should had been an alarm). Never from the
wrong
=
direction. Do you have the IGC files from a flight in which you remember
th=
at happening? It would be interesting to put it into SeeYou or other
softw=
are (or even look at it in a text editor) to see what the accuracy of fix
w=
as. I have noticed that the Flarm GPS is typically reporting a larger
error=
, and in some cases quite large. I'm not sure what the algorithms do with
t=
he precision of fix, but it seems like that is the most likely source of
th=
e errors you describe. For example the Flarm IGC file from my glider will
s=
how a typical precision of fix of around 3 - 4 meters, but sometimes it
wil=
l go up to 30-40 for unknown reasons. The Air Avionics gps will show a
prec=
ision of 1 - 2 meters on the same flight and might also go up in the same
a=
reas but not as much. Ridge flying in the Alps you might have the antenna
s=
haded on one or more sides, which will increase the HDOP. The precision
is
=
the normally the last three digits of the B record in the IGC file. All
of
=
my flying is high altitude and with a clear view of the sky.

"This is a fundamental flaw in Flarm, which surely could be much improved by
building in a wind algorithm to correct the indication nearer to heading.
All navigation programs have them."

As another poster mentioned, you really need a heading input to achieve this. Whilst PNA's use drift while circling amongst other methods to determine the wind, they require frequent & sustained 'circles' to achieve this - not so good for wave & ridge. LX quote 3 circles from memory and all the operating notes warn of the associated unreliability. An attempt to use an algorithm to achieve this in Flarm would result in large variations in accuracy. Sometimes the relative bearings provided would be correct and sometimes, they wouldn't. Though currently an imperfect system, at least it's consistent.

CJ

At 13:30 27 November 2015, wrote:
"This is a fundamental flaw in Flarm, which surely could be much improved
b=
y=20
building in a wind algorithm to correct the indication nearer to
heading.=
=20
All navigation programs have them."

As another poster mentioned, you really need a heading input to achieve
thi=
s. Whilst PNA's use drift while circling amongst other methods to
determin=
e the wind, they require frequent & sustained 'circles' to achieve this -
n=
ot so good for wave & ridge. LX quote 3 circles from memory and all the
op=
erating notes warn of the associated unreliability. An attempt to use an
a=
lgorithm to achieve this in Flarm would result in large variations in
accur=
acy. Sometimes the relative bearings provided would be correct and
sometim=
es, they wouldn't. Though currently an imperfect system, at least it's
con=
sistent.

CJ

I only mentioned wave to illustrate the extreme example. Some error is
always there unless you are flying straight up or down wind. It is far
from consistent, the amount depends on the wind speed and your angle to it,
two variables.
So whats wrong with flying 2/3 circles, I often circle in wave. In any case
Paolo Ventrafridda developed a method for LK8000, of flying S & L on one of
several headings for say 10 secs at a constant airspeed.
If the will is there, it can be done, and even if slightly imperfect, it
would be better than the present almost always wrong indication.
However another approach would be a way of manually putting the wind into
Flarm, using the vario/navigator readout, which I'm sure everyone flying
with Flarm, is equipped with as well.
Dave

On Friday, November 27, 2015 at 4:00:05 PM UTC, David Salmon wrote:
At 13:30 27 November 2015, wrote:
"This is a fundamental flaw in Flarm, which surely could be much improved
b=
y=20
building in a wind algorithm to correct the indication nearer to
heading.=
=20
All navigation programs have them."

As another poster mentioned, you really need a heading input to achieve
thi=
s. Whilst PNA's use drift while circling amongst other methods to
determin=
e the wind, they require frequent & sustained 'circles' to achieve this -
n=
ot so good for wave & ridge. LX quote 3 circles from memory and all the
op=
erating notes warn of the associated unreliability. An attempt to use an
a=
lgorithm to achieve this in Flarm would result in large variations in
accur=
acy. Sometimes the relative bearings provided would be correct and
sometim=
es, they wouldn't. Though currently an imperfect system, at least it's
con=
sistent.

CJ

I only mentioned wave to illustrate the extreme example. Some error is
always there unless you are flying straight up or down wind. It is far
from consistent, the amount depends on the wind speed and your angle to it,
two variables.
So whats wrong with flying 2/3 circles, I often circle in wave. In any case
Paolo Ventrafridda developed a method for LK8000, of flying S & L on one of
several headings for say 10 secs at a constant airspeed.
If the will is there, it can be done, and even if slightly imperfect, it
would be better than the present almost always wrong indication.
However another approach would be a way of manually putting the wind into
Flarm, using the vario/navigator readout, which I'm sure everyone flying
with Flarm, is equipped with as well.
Dave

My colleague corresponded with Flarm during our Scottish trial in 2007 about the possibility of correcting the track/heading difference by wind estimates from circling and they said they would look it it but never introduced it. One of the obvious issues is that the modes of flight during which this difference is most obvious (ridge and wave) are less likely to entail a lot of circling. Also in mountain ridge flying we are more likely to experience varying local winds.

I think that trying to do this would introduce too many uncertainties and different calculations between gliders.

Even without Flarm a pilot who can't figure out his track versus heading when ridge flying would be looking out the window in the wrong place for conflicting gliders.

John Galloway

On Friday, November 27, 2015 at 12:32:47 PM UTC-5, wrote:
On Friday, November 27, 2015 at 4:00:05 PM UTC, David Salmon wrote:
At 13:30 27 November 2015, wrote:
"This is a fundamental flaw in Flarm, which surely could be much improved
b=
y=20
building in a wind algorithm to correct the indication nearer to
heading.=
=20
All navigation programs have them."

As another poster mentioned, you really need a heading input to achieve
thi=
s. Whilst PNA's use drift while circling amongst other methods to
determin=
e the wind, they require frequent & sustained 'circles' to achieve this -
n=
ot so good for wave & ridge. LX quote 3 circles from memory and all the
op=
erating notes warn of the associated unreliability. An attempt to use an
a=
lgorithm to achieve this in Flarm would result in large variations in
accur=
acy. Sometimes the relative bearings provided would be correct and
sometim=
es, they wouldn't. Though currently an imperfect system, at least it's
con=
sistent.

CJ

I only mentioned wave to illustrate the extreme example. Some error is
always there unless you are flying straight up or down wind. It is far
from consistent, the amount depends on the wind speed and your angle to it,
two variables.
So whats wrong with flying 2/3 circles, I often circle in wave. In any case
Paolo Ventrafridda developed a method for LK8000, of flying S & L on one of
several headings for say 10 secs at a constant airspeed.
If the will is there, it can be done, and even if slightly imperfect, it
would be better than the present almost always wrong indication.
However another approach would be a way of manually putting the wind into
Flarm, using the vario/navigator readout, which I'm sure everyone flying
with Flarm, is equipped with as well.
Dave

My colleague corresponded with Flarm during our Scottish trial in 2007 about the possibility of correcting the track/heading difference by wind estimates from circling and they said they would look it it but never introduced it. One of the obvious issues is that the modes of flight during which this difference is most obvious (ridge and wave) are less likely to entail a lot of circling. Also in mountain ridge flying we are more likely to experience varying local winds.

I think that trying to do this would introduce too many uncertainties and different calculations between gliders.

Even without Flarm a pilot who can't figure out his track versus heading when ridge flying would be looking out the window in the wrong place for conflicting gliders.

John Galloway

From the flarm.com press release of 2015-01-28, in part: "...It includes safety features that increase the effectiveness and robustness of collision warnings, further decreasing nuisance alarms, for example by taking into account wind."

It also talks about the FLARM TrackingServer release "...in spring 2015...", which as far as I can tell, didn't happen.

It would be helpful if someone from FLARM could comment on how the wind is taken into account, and the status of the TrackingServer. I note it would be convenient if they had a forum/bulletin board where customers of their expensive and complex products could interact with them and each other.

James mentioned false alarms while diametrically opposed in thermals. That is highly unlikely to be due to wind drift.

On Fri, 27 Nov 2015 09:31:17 -0800, jfitch wrote:

James mentioned false alarms while diametrically opposed in thermals.
That is highly unlikely to be due to wind drift.

.... which is something I've never experienced, but maybe I've never
shared a thermal with an idiot since I've had FLARM fitted. That said, at
my club there was one collision in a thermal between two FLARM-equipped
gliders. AFAICT from talking to the pilots, one of them was far from
being on the diametrically opposite side of the thermal and then misread
the intentions of the other pilot. Under these conditions FLARM won't
help because the time between its warning being triggered and the
collision is likely to be too short for either pilot to do anything about
it.


--
martin@ | Martin Gregorie
gregorie. | Essex, UK
org |

On Friday, November 27, 2015 at 11:15:40 AM UTC-8, Martin Gregorie wrote:
On Fri, 27 Nov 2015 09:31:17 -0800, jfitch wrote:

James mentioned false alarms while diametrically opposed in thermals.
That is highly unlikely to be due to wind drift.

... which is something I've never experienced, but maybe I've never
shared a thermal with an idiot since I've had FLARM fitted. That said, at
my club there was one collision in a thermal between two FLARM-equipped
gliders. AFAICT from talking to the pilots, one of them was far from
being on the diametrically opposite side of the thermal and then misread
the intentions of the other pilot. Under these conditions FLARM won't
help because the time between its warning being triggered and the
collision is likely to be too short for either pilot to do anything about
it.


--
martin@ | Martin Gregorie
gregorie. | Essex, UK
org |

The short warning you get in thermals is a limitation of Flarm. Mitigated by a very good tactical screen of which there are unfortunately few examples.. One of the compromises that appear to have been made to eliminate false alarms in thermals is a very short warning distance. I have flown close to other gliders in thermals (yes they were aware) to see just when the alarms occur. I'm not criticizing Flarm for this, something I think they had to do.. Too many extraneous alarms is as bad as no alarms at all.

On a good tactical screen (the original Winpilot remains the very best by a wide margin) gives you a 3D map of all the gliders in the thermal near your altitude. Very easy at a glance to see where everyone is. I do not know of another display with this capability but I found it very informative. Unfortunately the original Winpilot doesn't work with modern equipment anymore, so the facility has been lost.

On Fri, 27 Nov 2015 11:27:06 -0800, jfitch wrote:

On Friday, November 27, 2015 at 11:15:40 AM UTC-8, Martin Gregorie
wrote:
On Fri, 27 Nov 2015 09:31:17 -0800, jfitch wrote:

James mentioned false alarms while diametrically opposed in thermals.
That is highly unlikely to be due to wind drift.

... which is something I've never experienced, but maybe I've never
shared a thermal with an idiot since I've had FLARM fitted. That said,
at my club there was one collision in a thermal between two
FLARM-equipped gliders. AFAICT from talking to the pilots, one of them
was far from being on the diametrically opposite side of the thermal
and then misread the intentions of the other pilot. Under these
conditions FLARM won't help because the time between its warning being
triggered and the collision is likely to be too short for either pilot
to do anything about it.


--
martin@ | Martin Gregorie gregorie. | Essex, UK org |

The short warning you get in thermals is a limitation of Flarm.
Mitigated by a very good tactical screen of which there are
unfortunately few examples. One of the compromises that appear to have
been made to eliminate false alarms in thermals is a very short warning
distance. I have flown close to other gliders in thermals (yes they were
aware) to see just when the alarms occur. I'm not criticizing Flarm for
this, something I think they had to do. Too many extraneous alarms is as
bad as no alarms at all.

Yes - agreed. I wasn't criticising FLARM at all for this. IMO the reality
of thermal gaggles is that everybody *must* maintain situational
awareness in a multiply occupied thermal. Thinking about it a bit
further, if everybody at more or less the same height in a thermal flies
sensibly, the time to go from safe to collision takes enough time to make
avoidance fairly easy. It would take at least one pilot to be grossly out
of position to make a collision imminent (think leaving by blasting
across the centre or getting too close behind or below another glider).
FLARM will spot a dangerous joining manoeuvre but whether the warning
would do you any good may depend on where the other glider(s) in the
thermal are, i.e. do you have an escape route that doesn't endanger
anybody else.


--
martin@ | Martin Gregorie
gregorie. | Essex, UK
org |

I am still seeing a lot of misinformation out there. I have two points to make supporting the use FLARM stealth mode in contests.

1) Stealth mode still allows the display and audio warning for threat aircraft no matter what the range.

and

2) FLARM used without stealth mode leads to an invalid score sheet. This is more true in eastern U.S. or European contests with lower working bands and more potential landouts.


First, I'd like folks to understand that FLARM sends two different messages to the display devices.

The $PFLAU sentence has priority and contains info about intruder alerts and obstacles. The contest ID is removed in stealth mode. Alerts are unaffected no matter the range. It really works quite well with the algorithm the FLARM people have developed.

The $PFLAA sentence is info about proximate aircraft displayed on your device. In stealth mode this info limited to aircraft within 2 km and +/- 300 meters vertically. Stealth or competition mode also removes ID, climb rate, track and speed from the display output for these proximate aircraft. It continues to use these variables to calculate the collision avoidance algorithm in $PFLAU.

Folks should read FLARM release notes for FLARM 6.02 Firmware, FLARM data port specification TFD-12 and FTD-14 FLARM Configuration Specification for full understanding. Anyway, we found in Elmira last year it worked quite well and the contest was definitely still fun for all.

High Western conditions versus lower Eastern (US) conditions: Without the use of stealth mode, in a contest with a lower working band, a pilot relying on FLARM technology can drive harder without fearing a landout, knowing there are gliders ahead to mark thermals. This does work in the east where thermals are closer together and you may be one thermal away from a landout. Even a mediocre pilot who might not even be able to get around the course by him/herself that day can use FLARM to pick the best thermals, found by others, and do fairly well on the score sheet. I agree in most cases this will not get a pilot the win. I do believe FLARM without stealth mode jumbles the middle of the score sheet and leads to an invalid result.

So, do what you want when flying cross countries at home. However, I go to contests to see how I am stacking up against some great pilots. Stealth mode (soon to have more appropriate name) is the way to go here. It retains all the safety features it was designed to deliver, keeps your eyes outside of the cockpit where they should be and at the end of the contest period the score sheet shows which pilots have the best soaring skills.

XC