CI bisect: does the camera resolution change alone break the gripper?#657
Draft
davetcoleman wants to merge 2 commits into
Draft
CI bisect: does the camera resolution change alone break the gripper?#657davetcoleman wants to merge 2 commits into
davetcoleman wants to merge 2 commits into
Conversation
…ive test progress Two CI-infra changes folded together: 1. Bump the reusable workflow ref to PickNikRobotics/moveit_pro_ci v0.3.1, set enable_gpu: true, and switch the runner label from picknik-16-amd64 to picknik-16-amd64-gpu. v0.3.1 appends the CUDA image suffix when enable_gpu is true (moveit_pro_ci#26) -- without it, v0.3.0 set the label but kept the non-CUDA image, so MuJoCo's EGL rendering still ran through llvmpipe on CPU. 2. Add src/lab_sim/test/conftest.py with two pytest hooks (logstart, logreport) that write directly to fd 2, bypassing pytest's --capture=fd. Without this, a CTest timeout kills pytest before any per-test output is flushed, leaving the CI log silent past 'collected N items'. The hooks emit START / PASSED|FAILED|SKIPPED + duration for every test, so the next timeout (if any) names the offending objective.
…obot configs
Sets `resolution="1280 720"` on every <camera> in the workspace's MJCF scene
files, and ensures every top-level scene's <global offwidth/offheight> is at
least 1280x720 so the offscreen render buffer fits the largest camera.
Previously inconsistent:
- 1280x720 (lunar_sim, hangar_sim, picknik_accessories/ur5e wrist_camera)
- 640x480 (lab_sim, kitchen_sim, grinding_sim, april_tag_sim, dual_arm_sim,
factory_sim, space_satellite_sim, kinova_sim and variants)
This was discovered when `lab_sim` failed to start with:
[ros2_control_node] [ERROR] The published robot description file (urdf)
seems not to be genuine. ... Camera resolution mismatch, set:
<global offwidth="1280" offheight="720"/> in the MJCF model.
The wrist_camera in picknik_accessories/mujoco_assets/ur5e/ur5e.xml was
already at 1280x720, but lab_sim's scene declared the global offscreen
buffer at only 640x480, so the hardware plugin rejected the URDF.
Standardizing to 1280x720 everywhere keeps the included ur5e wrist_camera
working in every UR-based config and avoids the same trap silently waiting
in the others.
space_satellite_sim_camera_cal kept at 1920x1080 (intentional higher
resolution for camera-intrinsics calibration data).
phoebe_sim (submodule) is not included; it needs a separate PR in
PickNikRobotics/phoebe_ws (and was already at 1280x720 anyway).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
Consider whether the change should land upstream in Overlapping files
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Not for merge — second bisection step for #648.
#655 showed that
9.3.0-rc9itself is fine (34 pass / 82 skip / 2 unrelated flakes with no scene changes). So the gripper-bringup failure in #648 must come from the scene XML changes.This PR is
main+ commit 3 from #648 (CI yaml + conftest, same as #655) + commit 2 from #648 (camera resolution standardization to 1280×720, plus the matching<global offwidth offheight>bumps that have to go with it). No timestep change.If this PR's CI fails the same way (every test times out at 180 s on gripper wait) → camera resolution is the trigger.
If this PR passes → keep going, bisect the timestep next.
The
lab_sim/description/scene.xmlcherry-pick had one conflict — commit 2 incidentally reformatted the<option>block from one line to multi-line because commit 1 had addedtimestep="0.003"to it. I resolved by keepingmain's single-line<option>(no timestep). The camera/offwidth changes — the actual point of commit 2 — landed cleanly.