New Trajectory Planner - Testers/programs wanted

More
12 years 8 months ago #43472 by skunkworks
Replied by skunkworks on topic New Trajectory Planner - Testers/programs wanted
This is v2.5.3-2611-g9ea88c1

I get this running steve.ngc

sam

rellenberg wrote: That's strange, steve.ngc seems to run fine in my simulation and RT builds. However, I ran a big batch of tests and found a small overage on arc-parabolic-blend.ngc (attached). Is it consistently going over at a particular spot?

Attachments:
More
12 years 8 months ago #43475 by skunkworks
Replied by skunkworks on topic New Trajectory Planner - Testers/programs wanted
here is the x axis - first at around line 25-27 and 27-28...

That is with steve.ngc strait G64

sam
Attachments:
More
12 years 8 months ago #43485 by Rick G
Rob,

I'm working on an update to push out to the beta branch that will fix this and a few other minor issues.


I will update what I am testing this week. I did notice that end moves might be treated differently in certain circumstances.
Again I may not have the latest but here is the current TP vs the new with straight G64.





Rick G
Attachments:
More
12 years 8 months ago #43492 by skunkworks
Replied by skunkworks on topic New Trajectory Planner - Testers/programs wanted
Rob - I added your program to the bug tracker here..

sourceforge.net/p/emc/bugs/344/

although - I don't know if it will get fixed before your branch would get added.. (as no one has come forward with having issues with the current TP. - that I know of anyway)

sam
More
12 years 8 months ago #43508 by rellenberg
Replied by rellenberg on topic New Trajectory Planner - Testers/programs wanted
Hi Rick,

Thanks for finding that missing blend in your test run. I figured out why it's doing that. I made a dumb optimization about a month ago that led to some parabolic blends not being detected correctly. I wanted to pre-calculate the peak velocity for each segment, but this meant every time something changed about the segment length, the calculation would have to be redone. I'm reverting to the original behavior because it's a big headache for too little gain.

Another mistake: I scaled back acceleration by 0.5 if the current or previous segment has a parabolic blend. However, the check for this in some cases happened before blend arcs were created, which meant some segments were accelerating too slowly. I'll have a fix for both of these issues out by the end of the day on the experimental branch.

Sam, I'm still trying to reproduce that overage on my configs, but if I can't, I'll download your config again and try it out.
More
12 years 8 months ago #43509 by skunkworks
Replied by skunkworks on topic New Trajectory Planner - Testers/programs wanted
Hmm - that is odd. I could try it on a different computer. That was a sim build on my laptop.

sam
Time to create page: 0.223 seconds
Powered by Kunena Forum