Flarm Range Analysis
On Tuesday, June 13, 2017 at 8:51:26 PM UTC-7, wrote:
I raised all of these issues with FLARM years ago. The typical statistics from a flight do not support the range analysis very well at all. I asked for the ability to upload multiple files but was completely blown off by FLARM. I wrote some code to add together many IGC files and it was quite instructive. Their tools work quite a bit better. The problem is that people really don't understand how to read these plots. What you are really getting is the mean range. The problem with that is that contact is really a statistical problem. Really there should be three range rings: Mean range, mean range plus one standard deviation, and mean range minus one standard deviation. I suggested this as well but was completely blown off again.
To get a minimum range with a reasonable contact probability from a safety standpoint, you really should look at the mean minus one standard deviation. Those are the conditions under which you would have a reasonable confidence interval and know that you will hav a good signal. I suspect that in a large majority of cases, this may actually be a negative number. Unfortunately, people rely on this tech to alert them to a potential collision and the reality is that there is a relatively high probability that it will not do so. People use anecdotal evidence to support unreasonably long contact distances, "the other day I saw XX on my FLARM display from 15km away!", so they assume that is the norm but it is really probably mean plus multiple standard deviations. Unfortunately, they think that is normal when it isn't.
FLARM is hampered by several technological problems. Low power transmitters, poorly placed and poorly performing antennas, and low power cpus with insufficient horsepower to handle lots of targets in close proximity. It will never work right.
ADS-B uses high power transmitters, reliable position reporting and good antennas that are well placed. ADS-B targets can be easily tracked from 50 miles out. For an anti collision system, I want something that will give me good advanced notice that something is nearby and be compatible with all the other air traffic because it doesn't matter whether I hit another glider or a power plane, it's going to hurt either way.
FLARM is a highly flawed product and I won't have it in my aircraft.
Brakes don't work very well, so I won't have them in my car. They take a long time to stop you, the stopping distance is a statistical average depending on conditions. Yet people actually depend on this flawed concept to keep them from hitting anything. They will never work right. Some people say, "yesterday a truck pulled in front of me but my brakes just stopped me in time!" not realizing that if the road were wet, they would not have. What we need is tractor beams from the Starship Enterprise. They are high power and would stop you instantly, you would never hit anything.
Rather than dis the Flarm range tool, why don't you write one that works the way you want? All the information is in the files and is easily parsed. It doesn't even seem like that challenging an app to write.
|