Help with new CRS A255 5DOF robot arm simulation

More
05 Apr 2013 09:54 #32295 by sholmes
Finally getting back to tinkering with LinuxCNC. I'm working towards integrating a CRS A255 5DOF robot arm.

My first step in this direction is to build a simulation using vismach.

I've worked through most of this using the puma560 as a template using genserkins essentially doing the following:

1) remove the 6th joint across the whole platform
2) configure the new DH paramters
3) configure the ini file with new min/max values for the joints
4) create, orient and articulate the new .obj files in the a255gui vismach script

It's not pretty yet, but it's starting to be functional. I can move the simulation in Joint Mode just fine. The problem I'm running in to now is that while I can run the puma560 in World Mode, I can't do the same with my a255 simulation and even though the commanded position changes, the actual position does not. Seems to me there's a problem coupling the commanded position back in to the inverse kinematics. I've compared against the puma560 files but so far am unable to find where the issue is.

When operating I can see the backplot drawing what I'm commanding, but the simulation view doesn't move.

Of course you'll have to do the following before running:
sudo cp a255gui /usr/bin/a255gui

I've attached my configuration for others to play with if interested. Of course it's incomplete and totally untested against the actual arm, so use at your own risk.

Appreciate any help I can get.
Attachments:

Please Log in or Create an account to join the conversation.

More
05 Apr 2013 23:37 #32313 by ArcEye
Hi

You may have seen these already, but in case not

www.linuxcnc.org/index.php/english/forum...xkins?start=40#28172

wiki.linuxcnc.org/cgi-bin/wiki.pl?Videos

www.linuxcnc.org/index.php/english/forum...robot?start=10#29577

Might give some contacts to scratch heads with if nothing else

regards

Please Log in or Create an account to join the conversation.

More
08 Apr 2013 10:33 #32414 by sholmes
Thanks ArcEye I had looked at those. I guess I was hoping those same personalities would chime in. No worries though I'll keep digging and post as I figure my way through this.

What I should have included is that I'm running on Vers. 2.5.2 (Sim) as I'm doing preliminary testing/development under a VMware instance. Initially I was trying to run off what had been included in the LiveCD (2.4.6?) but then understood that the realtime errors I was seeing could have been contributing to my issue. I've now moved over to 2.5.2 and am still experiencing the issue where I cannot apparently perform inverse kinematics (jogging in World Mode doesn't move my model even though the Axis backplot moves, then switching back to Joint Mode makes the Axis backplot jump back to the position defined by the joint positions).

I'm still having the issue, though what I've learned since then (and corrected/tested for) is:

1) Realtime errors may be a factor in this issue causing Genserkins to fail (anyone care to comment/confirm/contradict this?)
2) Genserkins appears to require 6 joints to be defined (no more, no less). Unused joints have to be configured as dummies. (Any known update to this, or is this still the case?) I've now added back a 6th joint.

These points were discussed in this message thread from 1 1/2 years ago.
www.mail-archive.com/emc-users@lists.sou...ge.net/msg30723.html

Please Log in or Create an account to join the conversation.

More
08 Apr 2013 10:34 #32415 by sholmes
And here're my current configurations.
Attachments:

Please Log in or Create an account to join the conversation.

More
09 Apr 2013 06:12 #32444 by andypugh

2) Genserkins appears to require 6 joints to be defined (no more, no less). Unused joints have to be configured as dummies. (Any known update to this, or is this still the case?) I've now added back a 6th joint.

Did it help?

Please Log in or Create an account to join the conversation.

More
09 Apr 2013 07:50 #32447 by sholmes
Andy,
Unfortunately not. I can see now that the D-H parameters are likely off. Manipulating the end effector (Joints 3&4) in Joint Mode give unexpected attitude changes in the Axis backplot.

I've found two contradictory sources for the D-H parameters so I'm working hard to bang my grey matter in to shape to try and derive the parameters myself.

For anyone who's interested, the geometry of the robot can be found in the user guide.

CRS A255 Robot User Guide

Please Log in or Create an account to join the conversation.

More
09 Apr 2013 15:57 #32464 by andypugh

I've found two contradictory sources for the D-H parameters so I'm working hard to bang my grey matter in to shape to try and derive the parameters myself.


There is a nice video linked from the Wikipedia page. However I have no idea if it is correct compared to the expectations of genserkins.

Please Log in or Create an account to join the conversation.

More
09 Apr 2013 22:50 #32488 by sholmes
Thanks Andy I really appreciate your help. As it turns out I've been periodically going back to that excellent video to help.

That was a question in my mind too if the results of that process work directly with genserkins, but from what I can see with the Puma560 example, they seem to.

One thing that's been confusing me however is the tool indicator in the Axis backplot.

Even with the Puma560, when I change the end effector pitch in Joint Mode (Joint 4), the cone of the tool indicator doesn't seem to change pitch as I would expect it to even as I watch the model move the tooltip doesn't seem to follow in the backplot. So this is adding to my confusion when I'm trying to validate my own kinematics under the a255 model.

Has anyone else noticed this?

Please Log in or Create an account to join the conversation.

More
09 Apr 2013 23:09 #32489 by andypugh

One thing that's been confusing me however is the tool indicator in the Axis backplot.


The Axis backplot only "knows" about the XYZABC values from the G-code.
It chooses to display those according to the [DISPLAY]GEOMETRY setting in the INI file.
The order in which you specify the GEOMETRY letters changes the representation in the backplot, but it might never be entirely correct.

Please Log in or Create an account to join the conversation.

More
09 Apr 2013 23:46 #32492 by sholmes
Cool, that may explain it, however if that's the case then why does it change orientation and move when in Joint Mode?

At the end of the day, if that's the case then it's the wrong tool for me to use to troubleshoot any kinematic issues. No worries... learning as I go!

Please Log in or Create an account to join the conversation.

Time to create page: 0.274 seconds
Powered by Kunena Forum