Skip to content

[Task Clean-up] Dexterous Part 8/8: Move the warp Direct task variants to isaaclab_tasks_experimental#6582

Draft
hujc7 wants to merge 4 commits into
isaac-sim:developfrom
hujc7:jichuanh/task-cleanup-dex-warp-exp
Draft

[Task Clean-up] Dexterous Part 8/8: Move the warp Direct task variants to isaaclab_tasks_experimental#6582
hujc7 wants to merge 4 commits into
isaac-sim:developfrom
hujc7:jichuanh/task-cleanup-dex-warp-exp

Conversation

@hujc7

@hujc7 hujc7 commented Jul 17, 2026

Copy link
Copy Markdown
Collaborator

Review Map

PR Status Depends on Exact changes
#6410 [Docs] Environment overview regen
#6411 Part 1/8: Newton cloner/cubric/visualizer fixes
#6412 Part 2/8: OVPhysX articulation + manager runtime
#6413 Part 3/8: Reorient Direct, torch
#6414 Part 4/8: MARL-to-single-agent fix + handover/camera Direct #6413 changes
#6418 Part 5/8: Reorient manager counterparts #6413 changes
#6421 Part 6/8: Handover + camera manager counterparts #6413, #6414, #6418 changes
#6415 Part 7/8: Benchmark success-rate utilities + docs #6413, #6414, #6418, #6421 changes
📌 #6582 Part 8/8: Warp variants → experimental (draft; merges last) (this PR) #6413 changes
#6324 [DO-NOT-MERGE] Lumped validation reference ALL

Summary

  • Moves the warp-first implementation of the Shadow reorientation and handover Direct environments (kernels, environments, kernel behavior tests) to isaaclab_tasks_experimental, following the torch-first mainline convention: torch task logic in isaaclab_tasks, warp implementations in the experimental track.
  • The variants register under -Direct-Warp-v0 IDs and reuse the stable configurations and agents from isaaclab_tasks.core. The Allegro warp variant keeps its existing legacy registration in the experimental package.
  • The diff is largely the review-polished warp implementation from the lumped validation branch [DO-NOT-MERGE][Task Clean-up] Dexterous: lumped validation branch (series reference) #6324, relocated with import-path updates.

Stacking

Validation

Add a behavioral Metrics/success_rate signal (goal-reach streaks per
episode) and threshold-independent episode orientation-error diagnostics
to the Direct reorientation environments, with the shared helpers in
isaaclab_tasks.core.utils and torch math tests. The task logic is
torch-first per the mainline convention; success gates task health while
reward stays diagnostic. Also fix hand resets that could initialize
joints below their lower position limits.
hujc7 added 2 commits July 18, 2026 04:03
Fold the reviewed lump changes that belong to this part's content:

- Share per-family sim settings through task-cfg base mixins.
- Deduplicate backend scene presets via inner-class defaults and nest
  single-consumer helper cfgs in the shadow-hand Direct cfg.
- Compute the orientation error through
  isaaclab.utils.math.quat_error_magnitude and delete the local
  direct_reorient_rotation_distance primitive.
- Rename direct_reorient_reward to reorient_reward: shared symbols
  carry no paradigm prefix.

Source commits on the lump branch: 2c22af0, 792e400,
42675b6, c6140f9, 173e9dc.
Move the warp implementation of the Shadow reorientation and handover
Direct environments (kernels, environments, kernel behavior tests) from
the mainline task package to isaaclab_tasks_experimental, following the
torch-first mainline convention. The variants register under
-Direct-Warp-v0 IDs and reuse the stable configurations and agents. The
Allegro warp variant keeps its existing legacy registration.
@hujc7
hujc7 force-pushed the jichuanh/task-cleanup-dex-warp-exp branch from 4ec232a to 2047011 Compare July 18, 2026 11:10
Mirror the mainline rename (reorient_reward) in the experimental warp
package: the launcher is family-scoped, not paradigm-scoped, and the
package already lives under direct/. Pure rename; no behavioral
change.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

isaac-lab Related to Isaac Lab team

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant