Hello ROS community,

as a long-standing partner of Mitsubishi in the field of robotics, we (Adiro Automatisierungstechnik) have developed a ROS 2 driver for Mitsubishi robots in close cooperation with our partner Software byTS GmbH.
Take a look at the driver on GitHub and give us your feedback:

Best regards
Nico

6 Likes

@Nico2691 I am getting a 404 error on the webpage.

We had to take the repo offline again temporarily at Mitsubishi’s request. We had used the Mitsubishi R3 protocol to activate the real-time interface, but Mitsubishi does not want to publish it and we only know about it because of our close partnership. We will take the part out and put the repo back online on Monday.
Sorry for the inconvenience.

Oh, that’s very sad :frowning:

So no realtime control of Mitsubishi… our motivation to test it is now instantly going down…

Link should be available again :slight_smile:

@destogl: We use the real time protocol.
It was just about the way the protocol is activated.
Now the melfa programm is loaded manually onto the controller.
Previously we had automated it. But it shouldn’t be too much effort this way either.

We look forward to your feedback

Thanks for the clarification! It is great to see more drivers that support high-bandwidth streaming commands and not simply low-bandwidth/long-distance intermittent commands. Do you have an idea of what streaming command rate the Mitsubishi controller and driver supports? 100 Hz? 500 Hz? 1 kHz?

As far as I know, the firmware on the Robot itself works with a maximum of ~285Hz.

1 Like

Yes, the cycle time for the new CR800 controllers is 3.5 ms. The older controllers (CR750) still run at 7 ms. The real-time interface is described in the Mitsubishi manual BFP-A3379-D.
You could even specify the setpoints in the form of motor pulses via the real-time interface.

1 Like

Hello
Thanks for sharing the repo! I have some question for that repo.

  1. Is this repo supporting the RV-2F-D robot(CR751-D controller)?
  2. I mentioned that the mitsubishi R3 protocol before, do you use the RT Toolbox pro software to activate the R3 protocol or use program to activate the R3 protocol?
    waiting for you reply,thank you very much.

The driver should also be compatible with your older controller version and your robot type. As far as I can see, nothing has changed in the real-time interface.

Regarding the R3 protocol: We used this in the first version of the repository to activate the real-time interface.
Due to a request from Mitsubishi, we no longer use it!
But you can create the programme for activating the real-time protocol manually with the Teaching Box or the Toolbox on the robot controller. It’s just a few lines and we have the example programme ready in the repository.

Just to make it clear:
Mitsubishi use the R3 protocol (rights are owned by Mitsubishi) to communicate with their robots (for example their programming software RT ToolBox3) This protocol can be used to transfer programs or other data. In other words, everything that is needed to program a robot using PC software. The protocol has been around for many years, but there is no official publication, which is why official use (as in our repository) is unfortunately not officially permitted.
However, this does not mean that Mitsubishi robots cannot be controlled via real-time stream. It only requires a manual intervention on the robot to switch it to real-time mode. Our original solution with the R3 protocol had just automated this manual intervention.

Hi it throws me the following problem when I try to build image. Am I doing something wrong?

Can you tell us please a bit more?

Which OS are you using?
How do you build the image?
Anything you can tell us.

Hello,

I managed to build image. However, now I have another problem, since I don’t have nvida graphics on my computer, I don’t know how to set Intel in ENV lines 9 and 10 in Dockerfile.


When I do docker compose up -d, he throws me this one
.

Otherwise I use Docker desktop on WIN 10.

Remove the GPU Part from the docker-compose.yml. From ros2-mitsubishi-novnc and from ros2-mitsubishi-moveit.
The ENV vars shouldn’t be a problem.

deploy:
  resources:
    reservations:
      devices:
        - driver: nvidia
          count: 1
          capabilities: [gpu]

We’ve updated the repository and split the Docker Compose files, so the GPU is no longer used by default.