topological identification of vertexes on a part

[freecad-heini-1]

Further information concerning topological naming:
https://forum.freecadweb.org/viewtopic. ... 80#p124880
http://www.jsoftware.us/vol7/jsw0703-27.pdf
https://www.freecadweb.org/wiki/Naming_project
http://www.ijcc.org/on-line2(pdf)/pdf/ijcc5-6.pdf
http://www.maruf.ca/files/caadoc/CAAMmr ... Naming.htm
https://forum.freecadweb.org/viewtopic. ... 054#p20054
https://web.archive.org/web/20150218170 ... jcc5-6.pdf

[microelly2]

Further information concerning topological naming:

Thank you for this colletion, will bew useful when I get my algorithm stable

[lambda]

@microelly2 it seems the forum ate a (rather lenghty) reply from me. In short: I will test you code when I have time.

I hope to rewrite my post eventually.

[triplus]

I just read this article:

http://www.sciencedirect.com/science/ar ... 00117#bib1

What i like about it is it gives a nice and clear overview. And it doesn't just focus on the proposed solution.

[saso]

I find it good to research different possibilities to solve this, however as I understand this is probably still the primary implementation of it that we have for now:

https://docs.google.com/document/d/1-d2 ... 2R9Ug/edit
https://github.com/ickby/FreeCAD_sf_master/tree/Naming

[triplus]

Options.jpg (47.41 KiB) Viewed 1093 times

I sincerely doubt current implementation doesn't have anything to do with one of this (or hybrid). Or there could be more to it and the graph isn't as complete as it could be. Therefore saying there is a better strategy compared to researching just isn't going to fly.

P.S. Image courtesy from the link above.

[triplus]

Compare.jpg (80.53 KiB) Viewed 1087 times

Looking at this table. It would be nice if our proposed solution could be placed in it. That is where do we stand with PID. Will it be Topology or Geometry based or hybrid. Is it heavily based on one of the existing methods (highly likely)? If yes/no what are the key points, advantages and disadvantages of the selected method we will use? I wouldn't mind if such answers could be provided at some point in the future. To remove the ambiguity involved to some extent (not only from the topology point of view).

[microelly2]

wow, a lot to read for tomorrow

I have improved my methods today a little bit

here the result

phpBB [video]

I want to look for a good topological match. When I have 5 good points pairs identified
than I can calculate from one part to the other using the coordinate system.
but I think topology is the most important thing:
Look at the crash test of a car: the only thing which will be invariant is the topology of some subparts.

the scripts and model is all in my nurbs wb.