MH900 axis configuration won't work

[ted-miller]

So I just discovered that the axis configuration on an MH900 is really weird, due to the extra-large power converters on the lower three axes.

============================================================
CONNECT(STO)                                        
        -- AXIS --  -- BRK --   -- CV  --           
    SV <123456789> <123456789> <123456789> ON_EN  OT
------------------------------------------------------------
R1 :#1 [456---123] [123456---] [111---234] ON_EN1 OT1   
============================================================

1. The axis-order is not one that will be recognized by Ros_CtrlGroup_ConvertSequentialJointOrderToMotoJointOrder
2. The MotoPlus API can't control the 9th axis in a single control group.

This could potentially be re-wired on the servo board to a supported configuration. But the local engineer in charge of MH900 wasn't immediately sure about the implications. More investigation would be required.

---

And while discussing this issue, it was mentioned that the GP4 also has a really weird axis configuration. I don't have details beyond "really weird".

[gavanderhoorn]

The MotoPlus API can't control the 9th axis in a single control group.

Should we not assert on this model until this has been fixed?

Just to prevent late user-facing errors?

And document the fact that model is unsupported.

---

Edit: same for GP4.

(I vaguely remember at least one user with a GP4, but can't find any reference to it, neither here nor in the beta1 repository. Perhaps MotoROS1?)

---

Edit 2: yes, ros-industrial/motoman#544 was posted by a user of motoman_driver (ie: MotoROS1) with a GP4.

[ted-miller]

I think that's a good idea. But we don't currently have true robot-model-detection. We just get the axis type (linear/rotational) and then make some assumptions based on number of axes.

Having said that, it's possible.

[ted-miller]

GP4:

I don't see anything weird here.

Perhaps due to the fact this is from MotoSim and not a real arm. But given that ros-industrial/motoman#544 got everything working, I'm guessing it's fine.

[gavanderhoorn]

But we don't currently have true robot-model-detection

any M+ APIs we could use?

[ted-miller]

Here's a couple HSES commands... I have to imagine there are M+ equivalents.

GetAxisConfiguration.pdf

SystemInformation.pdf

[ted-miller]

I suppose we could always just parse the PANELBOX.LOG file.
It's not ideal, but it wouldn't be too difficult.

[ted-miller]

Or we could invoke the HSES command. It's easy enough to hard-code a binary packet (as opposed to implementing an entire interface).

[gavanderhoorn]

Are we mostly worried about the active axis in the 9th slot, or about the unusual order?

[ted-miller]

Both equally.

We need to detect unusual order for the sake of Ros_CtrlGroup_ConvertSequentialJointOrderToMotoJointOrder. Otherwise we won't process an incoming trajectory properly.

We need to detect the 9th axis to raise a fatal alarm, since we can't control the U axis in that configuration.