3 close groups combined

[Nate The Surveyor]

This is a situation that happens occasionally. There are a number of solutions to it, that come to mind.
Bringing it up for discussion.
The obvious one is to enlarge the radius of the confidence guard. But, often, you don't realize the need, until you are already 10 or 15 minutes into an observation.
Let's discuss long term solutions.
N

 

[Sean Joyce]

Nate;
I will take first shot at it. How about a new white box with settings that identifies groups are clustered within a certain radius set by the user.
At first glance the box would list the number of groups within the tolerance (throwing out the ones that don't) that has been set.
The settings would be potentially confidence guard radius overlap between group setting, with a minimum amount of time required before this option is available?
If the user likes what he sees the position of those groups can be averaged, or if there is a known point that a particular group matches best than that group can be
accepted.

 

[Aaron S]

Is there currently any way to tell how close the groups are to each other? Based just on that screen, you can see that the max distance is 3.41, but how do you know those three clusters are "close enough" relative to each other in the first place?

 

[Nate The Surveyor]

Well, we don't know. But we have 3 groups, all with lots of time on them. In close proximity to each other. We guess, or assume they might overlap each other. Or, nearly overlap each other. While I was getting my phone out, one of the other groups picked up, and ran to 10. It then finished its cycle. It's just on my list of things... Maybe not top importance, but Javad has pushed the envelope. And, this has led to more ideas.
Ideas is where it starts.
N

 

[Sean Joyce]

Is there currently any way to tell how close the groups are to each other? Based just on that screen, you can see that the max distance is 3.41, but how do you know those three clusters are "close enough" relative to each other in the first place?

Aaron
The outlier groups outside of the tolerance would be thrown out and not used.
Would it be helpful if a table showing the coords of each group and relative accuracy between those groups was shown, minus the ones outside of the tolerance?

 

[Matt Johnson]

We guess, or assume they might overlap each other.

If you look at the vertical plot you can see that they do overlap as there are more epochs in most the groups than there are dots for each group so some of the epochs must be contained in multiple groups. I think that any method of trying to combine groups of groups over complicates things and is not neccessary. The settings for the size of the Confidence Guard are adjustable so if you want the groups to be larger, you can change this setting. If you want to use all epoch in final solution then select the PPK solution.

 

[Aaron S]

I'm not trying to crap all over the idea, but I'm also struggling to see why just increasing the size of the "buckets" wouldn't accomplish this.

If it stores all the hundreds of epochs from phase 1 (maybe in the background somewhere) for each point, I wonder how hard it would be to change the confidence guard after the fact for a point, and then have it recalculate. If it dumps all those epochs upon accepting the final shot, then that wouldn't work.