Hi,
Yes kdl.
Actually, we have some 6-R controller source c++ code, but with lots of bugs.
Aii. This sounds painfull.
Not developed by our team.
Wich team is your team, from point off interest?
Painful to repair and improve it
Clean coding is a skill that takes time. I think the previous team manager failed.
But coding at this math level is not an easy task at all. Solving bug's can take more time then write code
parts from scratch again.
Can we develop a 6-R robot control system based on the LinuxCNC ?
There are some around already. A few year's ago a professor made a script about linuxcnc with robot controller, it's still on the net.
And there are some other examples.
If you have the math level to do the inverse interpolation yourself in c code, it's quite easy to implement it in hal.
The advantage of kdl is that you can build up a structure from joint's or whetever what. I see no direct limit's to kdl.
But kdl is in c++ and c++ is not realtime. So you need a workaround.
But it is possible.
How to implement the command (such as MOVL, MOVC.....) parsing?
Trough hal you can sent command's like that to whatever part of lcnc. But you have to code the movl, movc parser yourself in c++.
I think you can keep the existing syntax. And write a simple translator for it. This should not be the problem.
