Roborealm + Kuka Robot
Victor Navratil from Romania  [3 posts]
7 months

My name is Victor, i am a student, in the final master year.
I (me and another colleague) have as a final project, to implement visual recognition, on a Kuka robot, with Roborealm software.
The robot is KR-15 with KRC2 controller.

Regarding this project, i have some questions:
1)which would be a better communication channel? ethernet or serial?
2)how can the robot retrieve the information sent? does he need some other software/addons
3)can he process information realtime?

Any ideas, suggestions, tips, would be of great help!
Thank you very much!
Have a great day, and sorry for my poor english!
Steven Gentner from United States  [260 posts] 7 months

1. Assuming you don't have any huge data requirements either Ethernet or Serial will be fine. I would use whatever you are more familiar with for now.

2. You'd have to become very familiar with the KRC2 controller and understand how information can be enter into the system remotely or can grab information remotely. We don't have a module that will communicate with a KRC2 controller directly mainly because we've not had access to the robot.

3. Realtime processing is relative to what you are trying to do. I.e. what you are asking is if processing can be done fast enough for the needed task. That depends on the actual task and how much processing is needed with how much processing power is available.

What I would first research is how you can control the robot either using Ethernet or Serial to be able to command it to move where you want it to. Once you understand how this works you can better understand how to work in RR to send those appropriate commands.

Victor Navratil from Romania  [3 posts] 7 months
Hello Steven,

Thank you very much for the information! Greatly appreciated!
It gave me clues on what to do next.
Have a great day!

Victor Navratil from Romania  [3 posts] 5 months

I've managed to create a bidirectional communication (serial) with the robot, as well as memorize the variables i need.

I would have a few more questions:
1) How can i limit the camera fps? I need 1 frame / 1-10 seconds.
2) Does RR have a "take picture" module? If yes, which is that?I would use this instead of limiting the frame rate.
3) Is there a function to stop the camera when not needed? Ex: when receiving a variable?
4) Can the transmission be stopped? Ex: when i set a variable

Thank you very much!
Have a great day!
Steven Gentner from United States  [260 posts] 5 months

1) There are several ways to do this. It is best done right from the camera itself. For example, if you are using a DirectInput camera go to the Options button->Video tab->Video Source/Format which will typically have a frame rate setting. This does NOT always work since these GUI interfaces are from the Camera manufacturer and NOT under our control ... but its worth a try.

The other method is to use Options Button->Other (last) tab and set the Max Processing to 1 which will force RR not to run the pipeline more than 1x a second. Despite the fact that the camera may be sending images RR will just ignore those frames.

The other method is to use the Timer module which will allow part of the pipeline to run at full speed but them limit the execution after the Timer module to the set time.

The other method is to use an IF Statement module and a timer variable to only execute the contents of the If Statement every X seconds.

2) Yes, its called Marker, or the ability to Mark/Name an image such that it can be referred to in other modules or brought back into the main processing pipeline. It basically always you to save the current image into memory and switch to another image.

3) Depends on the camera .. but this is typically not recommended since stopping the camera will cause a couple seconds delay on startup again which is typically not ideal since the event you wish to image may already have passed. More advanced Machine Vision cameras will offer a trigger that is accessible from the Genicam module ... so if you have one of those then that is also a possibility ... but its not present in generic webcams.

Normally in the machine vision world, one will "strobe" an image along with a flash of light. MV cameras have this as a connector on the camera. This strobing signal is a hardware one generated from a PLC or other timing device. This will cause the light to flash and the camera to snap an image. This image then becomes available to RR for processing. Normally in this mode the camera will appear locked since its not sending images so instead (esp when previewing) we do NOT set the camera in this mode and just use the light flash along with the Strobe RR module to trigger a processing event. This then uses the If Statement module to only execute a tab or contained modules when a flash is detected but otherwise ignores the mostly dark image.

In a nutshell, there are many ways to achieve what you want so please post with other questions/confusions.


This forum thread has been closed due to inactivity (more than 4 months) or number of replies (more than 50 messages). Please start a New Post and enter a new forum thread with the appropriate title.

 New Post   Forum Index