ROS LinuxCNC Link

Hi Arciera,

github.com/grotius-cnc/linuxcnc

I think it is oke now. I see multiple hal files appearing into the source.

Hi Arciera,

Did you had a look at the github code, if it compiles?
If not, i have to look at it.

At the moment finishing the mdi commands. Have to do the cartesian mdi.
What's left to do is moving in tool direction, Arciera's request.

I am really impressed by the realtime performance. Later on next week i will make some video's to show it.
Now also showing when joint limit is reached. Then the button color become's red. Implementing this failsafe logic
into the program is quite a lot of work.

For the controls i now have working :
1. joint move, j0,j1,j2,j3,j4,j5 with max deg a second.
2. cart move x,y,z with max lineair speed and acc-dcc in s-curve
3. tcp rotation in euler x,y,z with max deg a second.
4. joint limits
5. mdi joint command. The joint's will end at the same time.
6. live hal mode - demo mode
7. tools
8. realtime position stream in base-thread.
9. testing on ethercat hardware

When the controls are done and everything works how it should. I will integrate the pendant style again.

I got more and more the feeling this is a very good solution. It's just like streaming a netflix movie.



Most of the readers know we have managed to implement the dxfrw library this week with another program.
This break's open a way that we can load 3d dxf files into this program and for example generate a 3d cut path for the robot.
So we can go quite advanched soon !

Did you had a look at the github code, if it compiles?

I'm sorry, I have somehow missed your post. I'll try to compile again and report back.

Thanks

Ok, it has compiled successfully.

How do I now start your application?

Hi Arciera,

I need a few day's more to code. When it's ready to try will give you the download link.

Hi,

I made little progress. Added some technical details. It can look a bit messy, but for now it's ok.
The controls are fully functional. Now focussing on how to teach and how to store and edit this data in opencascade.

1.Moving the robot in tool direction is done! This works very nice. Also mdi commands are working.

2.When for example teaching a line, it will attach the toolposition's (euler angles) at the begin- and the end of the line to teach.
When clicking on the teached line, you can simple delete it (and it's childs) by one command. This is quite nice.

- The green cone is startpoint.
- The red cone is a halcommand including position & euler angles.
- The black cone is the endpoint.

3. Todo is getting the possibility to edit existing line point's to enable live preview. How the editing works, i have to ask at the opencascade help center. So far i haven't seen a solution how to do this.

So far so good.

Hi,

A tiny project update.

The lines, linestrips, arc's, circle's, splines, tool-io are now showing the euler angles. Oke this is nice.
I re-organized the opencascade function's.
Next i am planning to integrate the dxfrw library. Then we can load some dxf file's into the project and maybe
then we get some idea's how to program a dxf.




So far so good!

Hi,

Today i integrated the dxfrw library.

Now is the question how to program this dxf preview if we do a cut or glue operation.
If we load a 3d stepfile model, it would be nice if we can program this the same way.

We could perform a contour recognize algoritme with standard euler angles as output.
We could set the euler angles in tangential direction. This can be used for a cut angle correction when using plasma.

Maybe think some more before start coding.

Hi

A tiny update. It can now do a tiny program. I made a tiny video to show what is going on.

user-images.githubusercontent.com/448801...7d0-e38b01950fe4.mp4

If i turn the screencast off, the path following is perfect. At the end of line 4 you see it has difficulties. My pc is quite an oldy.
Motor's are connected. For info, this is realtime 1ms with scurve motion profile.

Hi,

Today i made a video sample where robot does euler angles.

user-images.githubusercontent.com/448801...8d6-e6319185ce63.mp4
user-images.githubusercontent.com/448801...f68-16cc9dcdd676.mp4

So far i am quite satisfied. At some points the c++ code has to improved a bit. It currently calculates the kinematics from init position.