[Question] Necessary procedure to import URDF files

[catachiii]

Description

Dear IsaacLab team. Thanks for providing us with up-to-date USD files for all kinds of humanoids. I have couple of questions about the necessary procedures to import the URDFs by myself.

Taking the URDFs of Unitree H1 for example (https://github.com/unitreerobotics/unitree_ros/tree/master/robots/h1_description/urdf), I have been trying to convert the given URDF to USD files with the How-to guide (https://isaac-sim.github.io/IsaacLab/source/how-to/import_new_asset.html).

I successfully converted the URDF with --merge-joints and --make-instanceable. However, when I'm loading the humanoid USD file similarly to the demos (https://github.com/isaac-sim/IsaacLab/blob/main/source/standalone/demos/bipeds.py). I encountered the issue that some of the visuals (like the logos, etc.) are not updated with physics, as shown below.

I'm not sure what I missed when converting. So far, all quadruped URDFs from Unitree work fine and similarly to what is provided from the Nucleus server.

Thanks for your help!

Jin

PS:
I also noticed there was a PR weeks ago importing Unitree G1 successfully. @abizovnuralem @Dhoeller19 would you mind sharing your experience with me?

[fooyuhui]

hello, can you try to use this USD file?
h1_with_hand.zip
I tested it, and it can work well. This way you can tell if there's a problem with the file or with the code.

[catachiii]

@fuyh20 It looks good to me although I think it is wrongly converted with fixed base, as the robot remains in the air.

From my side I think I didn't modify any code. I was just replacing the usd files and it seems not to work. That's why I was puzzled.

[catachiii]

Update:

I tried without --merge_joints, and now the visuals look all good. But the simulation can easily explode with the following error

2024-06-20 11:01:56 [57,157ms] [Error] [omni.physx.plugin] PhysX error: Illegal BroadPhaseUpdateData , FILE /builds/omniverse/physics/physx/source/lowlevelaabb/src/BpBroadPhaseABP.cpp, LINE 4023

[Lr-2002]

hi, have you solve the problem with illegal broadphase update data?