2 numbers, confidence, and phase1 to phase 2

Nate The Surveyor

Nate The Surveyor

Well-Known Member
The semi "magic numbers" we are looking for are 3/180
That's 3 independant fixes over 180 seconds of time. We could even argue/ discuss 2/180.
However, what do you think of allowing it to jump from phase 1, to phase 2, after it gets to either a user set-able "magic number"?

Maybe allow it to jump to phase 2, any time it has more than 1/2 of the confidence number, WITH over 180 seconds of time?
Would we loose anything?
I suggest it could save some time, at no loss of accuracy.
I'm just bringing this up for discussion.
Thanks!
Nate
 
Matt Johnson

Matt Johnson

Well-Known Member
5PLS
If you aren't concerned with the confidence number, why not just set it to a low number to begin with? I think this would just complicate things and add another setting that new users wouldn't understand.
 
Nate The Surveyor

Nate The Surveyor

Well-Known Member
It's not quite that simple.
IF I am sitting on a point for 30 seconds, THEN I want the confidence counter to HAVE to achieve a 10. Meaning, If I can achieve a 10 confidence, in 30 seconds, then I want it to then go to Phase 2.
This gives me pretty high confidence, that we are on the right group, or bucket.

HOWEVER,
IF we have sat on a point, for 5 minutes, and we have achieved 5 fixed solutions, over 200 seconds, then the TIME factor, is what can give us that same confidence, (or more) that the confidence counter would have given.

Meaning that the TIME factor can give confidence. So, we can now justifiably jump to phase 2. Then, at the end, one more reset. This would give the whole observation:
For exampleL
5 independent fixes, with a time separation, of 200 seconds. This gives time in place of the additional 5 independent fixes.
then, 120 epochs, without resets. (Mine is usually set to 120 epochs)
Then, at the end, one more reset.
Would this not yield a quite strong solution, with less time.

As it is, it CANNOT move up to phase 2 without 10 independent fixed solutions. Even if it takes 15 minutes.

Time in place of redundant independent fixes. (Time between independent fixes, = HIGH confidence).

LESS independent fixes are needed, when there is a long period of time, between 2 independent and reset fixes.

Or, to simplify it, Time = Confidence.




Nate
 
Matt Johnson

Matt Johnson

Well-Known Member
5PLS
A better and simpler method to handle this would be to revise how the Confidence Number is calculated so that time is weighted into it. We had discussed this some in the past. Revising it so the incremental increases in confidence are a product of the time would work well I think.

CLi : incremental increase in confidence
ECLi : existing incremental increase in confidence (1, 1.25, 1.5, 1.75, 2.0 and 2.5 for 1 to 6 fixed engines)
T : time in seconds from first fixed epoch in the group

CLi = ECLi * T / 20


So for example at 20 seconds 1 fixed engine would increase the confidence by 1, at 40 seconds by 2 and at 180 seconds by 9.
 
Darren Clemons

Darren Clemons

Well-Known Member
I like the way you're thinking Nate. I've thought and noticed this EXACT same thing. Granted, we are trying to tweek the best working GPS system in coverage, but I see the same things you do.

When in the open, we want the confidence to be reached so setting it to a low number doesn't work, but in heavy coverage it is very frustrating when we get to that "magic number" you're referring to of 2,180+ or something similar. Once there, I 100% agree I'd like to "force" the LS to go to phase 2 because I'm 99.99% sure I've got the coordinate I want. By continuing to try and work through resets in phase one, all you usually get for possibly the next 5/10/15 minutes are epochs in other buckets you KNOW are incorrect. Not an efficient use of time for sure.

By somehow "forcing" the process to move onto phase 2, then the next fix in the correct group will stay "locked on" and not manually get reset allowing the multiple epochs of phase 2 to "average" out the position. We still may not get through phase 3, but the additional epochs collected in phase 2 will produce a much more accurate coordinate.
Of course, we'd still be very susceptible to a "fail jump" during phase 2, but with Matt's terrific suggestion of putting so much weight in the confidence on time, even if it did a fail jump, it wouldn't take forever to get back to phase 2.

I actually mentioned to Matt over a year ago when I first started seeing this sort of thing that I'd love to have the ability to toggle the different buckets as well as manually forcing phase 1 to phase 2, but I completely understand that this would be a bit too much "power" to put in us, the users, hands. That would take away too much from what the LS is designed to do and be very dangerous in the hands of a novice.

As I said, we're discussing making changes to no doubt the best working system there is in canopy, but what Nate's brought up here is a significant issue as far as I'm concerned. When we get to 2,180 or 3,220 or better, I think we've accomplished a ton of confidence and are ready to move to the next phase.

The changes Matt's suggesting or something similar would be a tremendous improvement and time saver!!
 
Nate The Surveyor

Nate The Surveyor

Well-Known Member
Thanks Matt.
I think your way of addressing my problem, is better than mine.
Perhaps simpler too!

Now,
What about making "jump fails", to have a time factor as well, so that time between independent fixes, is factored in as well?
 
Nate The Surveyor

Nate The Surveyor

Well-Known Member
0804171211.jpg

Here is an example where it would save time.
 
Matt Johnson

Matt Johnson

Well-Known Member
5PLS
A ticket has already been created to improve the confidence number calculation. We also have had a ticket to improve the problems with jump fails. What I have suggested is that any time a fixed epoch occurs during Phase 2 that is outside 2X the confidence guard, the verify process is sent back to Phase 1 but confidence numbers and time are not reset like currently occurs.

Send your gifts and bribes to @Michael Stazhkov to get these things done sooner.
 
Darren Clemons

Darren Clemons

Well-Known Member
Matt Johnson said:
A ticket has already been created to improve the confidence number calculation. We also have had a ticket to improve the problems with jump fails. What I have suggested is that any time a fixed epoch occurs during Phase 2 that is outside 2X the confidence guard, the verify process is sent back to Phase 1 but confidence numbers and time are not reset like currently occurs.

Send your gifts and bribes to @Michael Stazhkov to get these things done sooner.
Click to expand...
Awesome! Thanks guys even before its out. Honestly, there's not many of us (at least who post here) that are seeing what and how this will help the way it will. It's only relative on the worst of the worst spots (maybe 5-10%of our locations) - but that's where we're spending the majority of our time now. If this does what I think it will....oh boy! Could potentially be another hour or two per day saved in certain areas/farms.
 
You must log in or register to reply here.
Top