ECM Teleoperation Issue After Upgrading to dVRK 2.4.0

[Stillrainy]

Hi,

After upgrading the dVRK software stack to version 2.4.0, we’ve encountered issues with ECM teleoperation, making it nearly unusable. The following problems have been observed:

• The roll joint (joint 4) is overly sensitive, requiring extremely careful MTM movements to avoid abrupt rotation.
• In some cases, the ECM is unable to move along a certain direction, even though it has not reached any joint limits.

We’ve previously used ECM teleoperation extensively on version 2.3.1 without any of these issues.

Additionally, the teleoperation scale configuration appears to be ignored. In the Qt GUI, the panel shows all scales defaulting to 0.2, regardless of the values set in the JSON configuration file.

[adeguet1]

Can you confirm your system JSON configuration file has the proper "base_frame" defined for your MTMs? See for example https://github.com/dvrk-config/dvrk_config_jhu/blob/4d25c94aa0fb1a4289eb28772200a681aced2b9b/jhu-daVinci/system-SUJ-ECM-MTMR-PSM1-MTML-PSM2-Teleop.json#L63. Note that we spent some time standardizing names in JSON files and they are all lower case with _. Upper case is for acronyms only.

You can check your configuration files using Python with the schema files: https://github.com/jhu-dvrk/sawIntuitiveResearchKit/tree/main/share/schemas#readme

The new system configuration wizard can also be used to create new system config files: https://dvrk.readthedocs.io/main/pages/configuration/generators.html#system-wizard-use

[Stillrainy]

Hi Adeguet,

Thank you for your reminder, you're absolutely right, that was exactly the issue. The problem was resolved after reverting to the correct transform values.

In our previous JSON configuration, we used the standard HRSV transform for both MTMs, consistent with the matrices in the JHU config.

For the new system configuration, I used the wizard (which is really helpful for new users!) and selected the HRSV option with the -30° pitch. It generated the following transform:

{
    "IO": "IO",
    "IO_file": "arm/sawRobotIO1394-MTMR-30593.json",
    "arm_file": "arm/MTMR-30593.json",
    "base_frame": {
        "reference_frame": "HRSV",
        "transform": [
            [-1.0, -6.1232339957367648e-17, 1.0605752387249069e-16, 0.0],
            [0.0, 0.86602540378443871, 0.49999999999999994, 0.40000000000000002],
            [-1.2246467991473532e-16, 0.49999999999999994, -0.86602540378443871, 0.47499999999999998],
            [0.0, 0.0, 0.0, 1.0]
        ]
    }
}

The main difference here is the x-translation was set to 0, which doesn't account for the side of the MTM. I assume the wizard currently doesn't distinguish between MTMR and MTML based on arm name.

Would it be helpful to add a user option to select the MTM side, or at least include a note in the wizard reminding users to manually adjust the translation based on the arm?

[adeguet1]

@brendanburkhart Does the wizard set a different base_frame on MTM left and MTM right?

If not, it could be a GUI option with the default set based on the arm's name. We should use a prefix match so MTMR, MTMR1, MTMR2... etc would be identified as right MTMRs.