Repository navigation
Description
Question
I am checking the physical dimensions of the official Factory/FORGE peg-insertion assets before comparing training results. I noticed a possible discrepancy between the configured hole diameter and the geometry in the USD asset.
Could someone confirm whether the interpretation below is correct, and what the recommended way to handle this is?
Observations
In IsaacLab v3.0.0, factory_tasks_cfg.py specifies:
factory_peg_8mm.usd:diameter = 0.007986factory_hole_8mm.usd:diameter = 0.0081
However, measurements of the straight cylindrical portions of the official USD meshes, excluding the entrance chamfer, give approximately:
| Dimension | Configuration | USD mesh |
|---|---|---|
| Peg diameter | 7.986 mm | 7.986 mm |
| Hole bore diameter | 8.1 mm | 9.0 mm |
This would give approximately 0.507 mm of radial clearance, compared with 0.057 mm if the bore were actually 8.1 mm.
The measurements concern the inner cylindrical bore, rather than the outer bounding box of the hole asset.
Asset versions checked
The original geometry measurements were performed on the official Isaac 4.5 asset bundle used with an IsaacLab v2.3.2 setup.
I also checked the official assets referenced by the default IsaacLab v3.0.0 Kit configuration, which points to the Isaac 6.2 asset bundle:
https://github.com/isaac-sim/IsaacLab/blob/v3.0.0/apps/isaaclab.python.kit#L276-L278
Assets:
- https://omniverse-content-production.s3-us-west-2.amazonaws.com/Assets/Isaac/6.2/Isaac/IsaacLab/Factory/factory_peg_8mm.usd
- https://omniverse-content-production.s3-us-west-2.amazonaws.com/Assets/Isaac/6.2/Isaac/IsaacLab/Factory/factory_hole_8mm.usd
Both files have exactly the same SHA256 hashes as the previously measured assets:
factory_peg_8mm.usd
b5040c8c639f3dc0731adfb6022c0d63383b7550354e413dc24e7417b600869d
factory_hole_8mm.usd
4bad48cde5a2e523454f9ed584023fae53d619a5ffb3d6ab08f29a6240014c28
Previous runtime SDF sampling in the older setup also indicated a bore diameter of approximately 9 mm, consistent with the mesh.
For v3.0.0 / Isaac Sim 6.2, I have verified the source configuration and asset-file identity only; I have not repeated the runtime collision measurements.
Questions
-
Is the approximately 9 mm bore intentional, with the
8.1 mmconfiguration value serving a different purpose, or is this an asset/configuration mismatch? -
My understanding is that changing the
diameterfield alone does not resize the USD mesh or its collision geometry. Is that correct? -
Which asset version and physical clearance should be used when reproducing the intended Factory/FORGE peg-insertion benchmark?
-
If a different bore diameter is required, what is the recommended procedure for updating the USD geometry and regenerating its SDF collision representation?
I would appreciate confirmation of these measurements or guidance on any unit, transform, or collision-processing detail that I may have overlooked.
Build Info
- IsaacLab versions inspected: v2.3.2 and v3.0.0
- Official asset bundles compared: Isaac 4.5 and Isaac 6.2
- Runtime SDF measurements: older setup only
- IsaacLab v3.0.0 / Isaac Sim 6.2 runtime collision behavior: not tested
Activity
RandomOakForest commentedon Oct 28, 2026
Thank you for posting this. The team is actually working on a newer installment of such assets (see, this defs. for ref.). Would it be possible to use these instead of the legacy ones?