Robotic Linear Rail Build Photos

Hi thefabricator03
I want one of those too !!!! what a good job, I tried to make a similarity last year, but I stayed in the cinematics! My head doesn't reach that much! I'm looking forward to see how it works!
Regards
Chimeno

Roiki wrote:

How are you linking the tracks? are the linear rails 6m each or how are you aligning and joining them? And what a re you welding with 18m of track?

And yes running that much cable for each motor and encoder tends to cost a lot. Though signaling isn't an issue when handled right.

What I was thinking to join the tracks is I will have the existing rails cut at say 300mm and 150mm staggered in from one end and make the next set overlap onto the next track. I will have all the ends ground for the butt join like I did for this rail as I could only get 4000mm lengths max.

I will be using it to weld from 6m to 18m long web trusses,

chimeno wrote:

Hi thefabricator03
I want one of those too !!!! what a good job, I tried to make a similarity last year, but I stayed in the cinematics! My head doesn't reach that much! I'm looking forward to see how it works!
Regards
Chimeno

Thanks Chimeno,

The rail you designed looks interesting.

My strengths are my fabrication skills, I can make anything I put my mind to. But learning to code is not my strongest point. I am working on it slowly. I hope to by this time next year have a working six axis robot controlled by LinuxCNC mounted on top of my track and welding a webbed truss.
I will be using RoboDK as my CAM software.

I will keep everyone on this forum updated with my progress.

Cool. Aligning linear rails can be tricky though since they don't really like gaps in the rail. Hope it works out.

Ive also looked at roboDK since it's pretty much the only one available for hobbyists and I've been working on getting it working with linuxcnc. There are a few issues that I think you should talk to them about integration before buying it.

Roiki wrote:

.

Ive also looked at roboDK since it's pretty much the only one available for hobbyists and I've been working on getting it working with linuxcnc. There are a few issues that I think you should talk to them about integration before buying it.

What are the issue you have been having? I have spoken to Albert and he told me he would write a post processor for LinuxCNC when I buy a license.

I also wrote a post processor for linuxcnc and couldnt find a way to separate clockwise arcs from counter-clockwise arcs to get a proper g02/03 assignment going since no other robot code distinguishes them. Also I couldn't find a way to export trajectories as series of points that could be fed to linuxcnc without kinematics.

The big issue with linuxcnc and robots is that the kinematics module can't do any pose selection or singularity avoidance so inverse kinematics are useless. The only plausible way I've found is to only use forward kinematics and absolute joint values to drive a robot reliably. Driving a robotic arm is a lot more complex than what linuxcnc can handle really.

Cool build. Can you share how you were able to align the linear rails to prevent binding? I have only ever done this over short distances and found keeping things planer to be a real challenge.

I had a set of surplus thk rails that had the ends ground at a 45 for a "splice" overlap to extend them. Although a thk docs show a straight joint.

the bars in the First Image are supost to go the other way as they handle then alot more stiffness
good system will it be on Chain ,Timingbelt,RackPinion ???

Roiki wrote:

I also wrote a post processor for linuxcnc and couldnt find a way to separate clockwise arcs from counter-clockwise arcs to get a proper g02/03 assignment going since no other robot code distinguishes them. Also I couldn't find a way to export trajectories as series of points that could be fed to linuxcnc without kinematics.

The big issue with linuxcnc and robots is that the kinematics module can't do any pose selection or singularity avoidance so inverse kinematics are useless. The only plausible way I've found is to only use forward kinematics and absolute joint values to drive a robot reliably. Driving a robotic arm is a lot more complex than what linuxcnc can handle really.

Yeah I have known that there are things that LinuxCNC does not do yet like computed torque control of joints. Thank you for letting me know the other issues. I will look into what can be done about them. I have had the idea of bringing in the OROCOS kinematic and dynamics library into LinuxCNC but I was not sure if that would be useful as RoboDK should be sorting out the singularities and to a extent inverse kinematics.

So in your opinion LinuxCNC has to be able to handle the full inverse kinematics, dynamics and singularity avoidance?

OT-CNC wrote:

Cool build. Can you share how you were able to align the linear rails to prevent binding? I have only ever done this over short distances and found keeping things planer to be a real challenge.

I had a set of surplus thk rails that had the ends ground at a 45 for a "splice" overlap to extend them. Although a thk docs show a straight joint.

I did not do anything special with aligning the rails, I had the carriage set up on the rails when I was aligning them so I could slide it over after I put a bolt in. Then I just kept measuring the spacing as I went and kept checking with the carriage.

I drilled and tapped the rails as they were sitting in position. I would but a bolt in every 500mm as I went along until I had done the full length then I drilled and tapped the rest.