Anti Collision Warning

(Ramy Yanetz) wrote in message

These devices should be capable to tell you if you are in a collision
course, not just warn you of a nearby aircraft. Earlier someone posted
a link to a new device which also calculate collision course while
thermaling! If you thermal too close to someone else it should warn
you.


But how close is *too close*? I am perfectly comfortable cranked up
at a 50 deg bank with someone opposite me doing the same thing, but
very uncomfortable if another glider joins with the same separation
and puts me in their blind spot.

To be effective in providing warnings the device would have to
continuously predict collisions based not only on the current
trajectory of each aircraft, but also predict collisions based on all
possible future trajectories for the next say 30 seconds. Try
resolving that mess when there are 30+ gliders at the top of the same
thermal waiting for a contest start. The false alarm rate would be
unacceptable.

The nearest the US regulators have come to such a system is the
proposed ADS-B, but that seems to be a long way from the dream of 10
years ago that a GPS based ADS-B could be small and cheap enough to
be carried by ultralights and skydivers.

Andy (GY)

Andy Durbin wrote:
(Ramy Yanetz) wrote in message

These devices should be capable to tell you if you are in a collision
course, not just warn you of a nearby aircraft. Earlier someone posted
a link to a new device which also calculate collision course while
thermaling! If you thermal too close to someone else it should warn
you.



But how close is *too close*? I am perfectly comfortable cranked up
at a 50 deg bank with someone opposite me doing the same thing, but
very uncomfortable if another glider joins with the same separation
and puts me in their blind spot.

To be effective in providing warnings the device would have to
continuously predict collisions based not only on the current
trajectory of each aircraft, but also predict collisions based on all
possible future trajectories for the next say 30 seconds. Try
resolving that mess when there are 30+ gliders at the top of the same
thermal waiting for a contest start. The false alarm rate would be
unacceptable.

I think Andy is right, but I don't think that situation is the one that
produces the most collisions. My undocumented impression is the majority
involve just a few gliders, often just two. The computations for two or
three gliders should be easy compared to 6 or more. During thermalling
or beating back and forth on a ridge, gliders don't change relative
altitude very much, so this much reduces the potential paths.

So, a system that worked for 2 or 3 gliders would be useful, and as
experience was gained with it, I think it would be continually upgraded
to cover situations with more gliders. Even if it didn't work for more
than even 5 gliders, that would cover most situations.
--
Change "netto" to "net" to email me directly

Eric Greenwell
Washington State
USA

Of course the problem is which 5 out of the 7 in a particular thermal do you
track?

:-)

Ian

"Eric Greenwell" wrote in message
...
Andy Durbin wrote:
(Ramy Yanetz) wrote in message

These devices should be capable to tell you if you are in a collision
course, not just warn you of a nearby aircraft. Earlier someone posted
a link to a new device which also calculate collision course while
thermaling! If you thermal too close to someone else it should warn
you.



But how close is *too close*? I am perfectly comfortable cranked up
at a 50 deg bank with someone opposite me doing the same thing, but
very uncomfortable if another glider joins with the same separation
and puts me in their blind spot.

To be effective in providing warnings the device would have to
continuously predict collisions based not only on the current
trajectory of each aircraft, but also predict collisions based on all
possible future trajectories for the next say 30 seconds. Try
resolving that mess when there are 30+ gliders at the top of the same
thermal waiting for a contest start. The false alarm rate would be
unacceptable.

I think Andy is right, but I don't think that situation is the one that
produces the most collisions. My undocumented impression is the majority
involve just a few gliders, often just two. The computations for two or
three gliders should be easy compared to 6 or more. During thermalling
or beating back and forth on a ridge, gliders don't change relative
altitude very much, so this much reduces the potential paths.

So, a system that worked for 2 or 3 gliders would be useful, and as
experience was gained with it, I think it would be continually upgraded
to cover situations with more gliders. Even if it didn't work for more
than even 5 gliders, that would cover most situations.
--
Change "netto" to "net" to email me directly

Eric Greenwell
Washington State
USA

"Ramy Yanetz" wrote in message .com...
See my other reply below. These devices should be capable to warn you of
imminent collision even while thermaling, not just warn you of nearby
aircraft. And if they currently don't, they will in the near future. It
should not be too difficult to compute collision courses.


That is exactly what FLARM does.

"Eric Greenwell" wrote in message
...
Andy Durbin wrote:
(Ramy Yanetz) wrote in message

These devices should be capable to tell you if you are in a collision
course, not just warn you of a nearby aircraft. Earlier someone posted
a link to a new device which also calculate collision course while
thermaling! If you thermal too close to someone else it should warn
you.



But how close is *too close*? I am perfectly comfortable cranked up
at a 50 deg bank with someone opposite me doing the same thing, but
very uncomfortable if another glider joins with the same separation
and puts me in their blind spot.

To be effective in providing warnings the device would have to
continuously predict collisions based not only on the current
trajectory of each aircraft, but also predict collisions based on all
possible future trajectories for the next say 30 seconds. Try
resolving that mess when there are 30+ gliders at the top of the same
thermal waiting for a contest start. The false alarm rate would be
unacceptable.

I think Andy is right, but I don't think that situation is the one that
produces the most collisions. My undocumented impression is the majority
involve just a few gliders, often just two. The computations for two or
three gliders should be easy compared to 6 or more. During thermalling
or beating back and forth on a ridge, gliders don't change relative
altitude very much, so this much reduces the potential paths.

So, a system that worked for 2 or 3 gliders would be useful, and as
experience was gained with it, I think it would be continually upgraded
to cover situations with more gliders. Even if it didn't work for more
than even 5 gliders, that would cover most situations.
--
Change "netto" to "net" to email me directly

This still leaves the problem I think Andy was getting at of what is the
acceptable false positive:false negative ratio?
Too many false positives and pilots won't use it.
False negatives would lead to collisions, deaths and, at least in the US,
lawsuits that would likely put the manufacturer out of business.

Brent

Whatever happened to teaching good look out and airmanship?

Good for you! That's the best contribution to this thread I've seen.
David Starer

"Dave Martin" wrote in message
...
Whatever happened to teaching good look out and airmanship?

Good for you! That's the best contribution to this thread I've seen.
David Starer

"Dave Martin" wrote in message
...
Whatever happened to teaching good look out and airmanship?

"303pilot" brentUNDERSCOREsullivanATbmcDOTcom wrote in message
...
"Eric Greenwell" wrote in message
...
Andy Durbin wrote:
(Ramy Yanetz) wrote in message

These devices should be capable to tell you if you are in a collision
course, not just warn you of a nearby aircraft. Earlier someone posted
a link to a new device which also calculate collision course while
thermaling! If you thermal too close to someone else it should warn
you.



But how close is *too close*? I am perfectly comfortable cranked up
at a 50 deg bank with someone opposite me doing the same thing, but
very uncomfortable if another glider joins with the same separation
and puts me in their blind spot.

To be effective in providing warnings the device would have to
continuously predict collisions based not only on the current
trajectory of each aircraft, but also predict collisions based on all
possible future trajectories for the next say 30 seconds. Try
resolving that mess when there are 30+ gliders at the top of the same
thermal waiting for a contest start. The false alarm rate would be
unacceptable.

I think Andy is right, but I don't think that situation is the one that
produces the most collisions. My undocumented impression is the majority
involve just a few gliders, often just two. The computations for two or
three gliders should be easy compared to 6 or more. During thermalling
or beating back and forth on a ridge, gliders don't change relative
altitude very much, so this much reduces the potential paths.

So, a system that worked for 2 or 3 gliders would be useful, and as
experience was gained with it, I think it would be continually upgraded
to cover situations with more gliders. Even if it didn't work for more
than even 5 gliders, that would cover most situations.
--
Change "netto" to "net" to email me directly

This still leaves the problem I think Andy was getting at of what is the
acceptable false positive:false negative ratio?
Too many false positives and pilots won't use it.
False negatives would lead to collisions, deaths and, at least in the US,
lawsuits that would likely put the manufacturer out of business.

Brent



We can't ask for perfection or nothing will ever be available. It seems to
me that there are two indications we should be looking for in a basic
anti-collision device.

1, There are (n) gliders in close proximity - say 1 kilometer. Even simple
GPS broadcast devices should be able to determine the number of gliders
nearby. It should beep softly when the number changes. (If the device says
there are 3 gliders nearby and you can only see 2, you need to keep
looking.)

2, If one of these represents a collision danger, the device should give a
bearing. The device need only determine that the target is at or near the
same altitude, the distance is closing and the relative bearing is nearly
constant. If there is only 1% chance of an actual collision, that would get
my undivided attention.

Of course, collisions can occur with different geometry's but they are rare.

The device doesn't need to exactly predict a collision, only that one is
possible. That would be enough to get a pilots attention. There are a
couple a traffic alert/collision avoidance devices on the market now and
they don't seem too worried about legal troubles.

Commercial ADS-B devices will not be available for many years. The avionics
industry has to sell everybody a Mode S transponder first (Big$), THEN we
will get to upgrade to ADS-B(Still more big $). I hope these small efforts
within the soaring community produce something we all can use. We sure need
it.

Bill Daniels

"Dave Martin" wrote in message
...
Whatever happened to teaching good look out and airmanship?

Long and sad experience says it doesn't always work. "Just look outside"
is often an excuse for doing nothing about the problem.

Bill Daniels