MyoArm with no hand

[jamesheald]

These xml files are needed for PR 232.

Following Vittorio's advice, I have created a myoArm model that has no hand or hand muscles, for use in a pure reaching task.

I have created two new files:

1. myoarm_nohand_body.xml is a modification of myoarm_body.xml. All I have done is i) comment out the lines we don't need, and ii) added a new site at the end of the forearm called forearm_tip, which will act as the end effector in the reaching task.
2. myoarm_nohand_assets.xml is a modification of myoarm_assets.xml. All I have done is comment out the lines we don't need.

There is currently redundancy, as the arm-with-no-hand model is contained within the full arm model. A cleaner solution would be to define the full arm model as the arm-with-no-hand model with the hand included. However, as a first pass, I thought I would add these files rather than modify the full arm model.

[Vittorio-Caggiano]

@jamesheald can you remove the comments? also, can you add a snapshot with the description of # muscles | # joints for the Readme?

[Vittorio-Caggiano]

let's remove the commented parrs

[jamesheald]

@Vittorio-Caggiano I'll create a screenshot and a short description

[jamesheald]

I've done this. See the latest file changes to PR 232. I have added two descriptions/images, one for the model and one for the task.

[vikashplus]

There is currently redundancy, as the arm-with-no-hand model is contained within the full arm model. A cleaner solution would be to define the full arm model as the arm-with-no-hand model with the hand included. However, as a first pass, I thought I would add these files rather than modify the full-arm model.

+1 on this point. We are creating a redundancy that will be hard to maintain or keep in sync. We should explore ideas before committing to this path.

[Vittorio-Caggiano]

There is currently redundancy, as the arm-with-no-hand model is contained within the full arm model. A cleaner solution would be to define the full arm model as the arm-with-no-hand model with the hand included. However, as a first pass, I thought I would add these files rather than modify the full-arm model.

+1 on this point. We are creating a redundancy that will be hard to maintain or keep in sync. We should explore ideas before committing to this path.

indeed. We are temporarily accepting redundancy (hand replicated in arm). We need a more comprehensive long term approach

[vikashplus]

I have advocated for an attach functionality that is live and stable now. Perhaps it is time to look into adopting these - https://mujoco.readthedocs.io/en/stable/XMLreference.html#body-attach

We need more clarity on

• Attach assumes clean separation. Are hands and arms cleanly separable? In other words, are there any tendons that span both hands and arms?
• If that is not the case, we should look into mjSpec to decompose full arm model. We will need to detach arm from the hand in this case.

[Vittorio-Caggiano]

I do not think it is possible to completely separate arm from forearm ... we might need to decompose it

[vikashplus]

In that case, we need to move to using mjSpec for model editing. We will have to bump MuJoCo version for this feature.