-
Notifications
You must be signed in to change notification settings - Fork 1.3k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Update model files: for iiwa.urdf, use cylinders instead of meshes (for speed). #3507
Conversation
I'm very curious to have latency and memory benchmark numbers to see how big of different this makes! |
Thanks. I'm trying to make sense of these numbers. Does "#X" in the first column mean "body X" has a collision mesh? |
There are 8 main segments in the iiwa arm, representing the segments you can see in the arm. (Is "segment" a good term to use? Like the piece of arm between the wrist and elbow). Segments are links in urdf and are numbered 0-7. So when a segment is listed in my table above, it means only those links have meshes, and the remainder have cylinders. Review status: 0 of 15 files reviewed at latest revision, all discussions resolved, some commit checks failed. Comments from Reviewable |
Thanks. There appears to be almost an order of magnitude difference! 😄 |
Liang and I have gone over these results, which show that xdot from both the rbs and rbp have changed from the original iiwa model, but they also don’t agree with each other. rigid_body_plant_test results: urdf from this PR @amcastro-tri : Liang claims that the same urdf used in both a rbs and rbp should produce the same xdot values, and so there must be a bug in rbp. Can you please take a look? Review status: 0 of 15 files reviewed at latest revision, all discussions resolved, some commit checks failed. Comments from Reviewable |
Could the problem be that |
Yes, that could be a problem even if there are no collisions because a joint could reach a limit in which case Review status: 0 of 15 files reviewed at latest revision, all discussions resolved, some commit checks failed. Comments from Reviewable |
Change rbp damping to 15, to match rbs: Review status: 0 of 15 files reviewed at latest revision, all discussions resolved, some commit checks failed. Comments from Reviewable |
Had to match friction also, now results match and test passes. Review status: 0 of 15 files reviewed at latest revision, all discussions resolved, some commit checks failed. Comments from Reviewable |
Some build failures happened due to a dreal build failure, fixed in #3524. @drake-jenkins-bot retest this please Review status: 0 of 16 files reviewed at latest revision, all discussions resolved, some commit checks failed. Comments from Reviewable |
+@liangfok for feature review, please. Review status: 0 of 16 files reviewed at latest revision, all discussions resolved. Comments from Reviewable |
I see numerous Reviewed 15 of 15 files at r1, 1 of 1 files at r2. drake/examples/kuka_iiwa_arm/urdf/iiwa14.urdf, line 3 at r2 (raw file):
BTW: nit: "urdf" should be "URDF" since it is an acronym. (same comment below) drake/examples/kuka_iiwa_arm/urdf/iiwa14.urdf, line 11 at r2 (raw file):
Is "hinder" the right word here? I believe the purpose for having more accurate collision models for the end of the arm is because we anticipate that the end of the arm is where it will most likely contact the environment. Comments from Reviewable |
The dae's are for use in Gazebo, which apparently doesn't accept obj files. To use the URFD in Gazebo I did search and replace in the URDF of obj with dea. I don't want to create another URDF that will get out of sync. I could possibly remove the dae files if it's weird to have them here, unused. Regenerating them requires a tool like Meshlab. Review status: all files reviewed at latest revision, 2 unresolved discussions. drake/examples/kuka_iiwa_arm/urdf/iiwa14.urdf, line 3 at r2 (raw file):
|
I think being able to load and simulate the IIWA model in Gazebo is useful since it could be used as a comparison with Drake and to help debug and improve the model. I created #3543 to track this issue. The
Thus, we may need to address #3257 first. Review status: all files reviewed at latest revision, 2 unresolved discussions. Comments from Reviewable |
Removed all dae files in this example. Review status: 2 of 4 files reviewed at latest revision, 2 unresolved discussions. drake/examples/kuka_iiwa_arm/urdf/iiwa14.urdf, line 3 at r2 (raw file):
|
Leave meshes for links at end of arm, for model accuracy. Changed colors to better match reality. In our arm, only a part of the link is orange, and this colors the entire link orange, but its an improvement to see each link. Adding model.config for future work in Gazebo. Make friction coefficient and damping in rigid body plant match rbs.
Squashed commits so dae files do not appear in history. I think I did that right, anyway. Review status: 2 of 4 files reviewed at latest revision, 2 unresolved discussions. Comments from Reviewable |
Reviewed 2 of 14 files at r3. Comments from Reviewable |
+@jwnimmer-tri for platform review, please. Review status: all files reviewed at latest revision, all discussions resolved. Comments from Reviewable |
Reviewed 2 of 15 files at r1, 2 of 14 files at r3. drake/systems/plants/rigid_body_plant/rigid_body_plant.h, line 163 at r4 (raw file):
I am skeptical of changing magic numbers in a PR titled "update model files", with no apparent test coverage or justification. I see there were discussions of this in the comment threads, but is this change truly appropriate for this PR? Comments from Reviewable |
I left some more comments. Review status: all files reviewed at latest revision, 3 unresolved discussions. drake/examples/kuka_iiwa_arm/urdf/iiwa14.urdf, line 3 at r4 (raw file):
BTW: nit: replace "submitted" with "added". drake/examples/kuka_iiwa_arm/urdf/iiwa14.urdf, line 4 at r4 (raw file):
BTW: nit: insert back-ticks around drake/systems/plants/rigid_body_plant/rigid_body_plant.h, line 163 at r4 (raw file):
|
drake/systems/plants/rigid_body_plant/rigid_body_plant.h, line 163 at r4 (raw file):
|
Review status: all files reviewed at latest revision, 3 unresolved discussions. drake/systems/plants/rigid_body_plant/rigid_body_plant.h, line 163 at r4 (raw file):
|
Review status: all files reviewed at latest revision, 3 unresolved discussions. drake/systems/plants/rigid_body_plant/rigid_body_plant.h, line 163 at r4 (raw file):
|
Review status: all files reviewed at latest revision, 3 unresolved discussions. drake/systems/plants/rigid_body_plant/rigid_body_plant.h, line 163 at r4 (raw file):
|
Review status: all files reviewed at latest revision, 3 unresolved discussions. drake/systems/plants/rigid_body_plant/rigid_body_plant.h, line 163 at r4 (raw file):
|
drake/systems/plants/rigid_body_plant/rigid_body_plant.h, line 163 at r4 (raw file):
|
drake/systems/plants/rigid_body_plant/rigid_body_plant.h, line 163 at r4 (raw file):
|
Review status: all files reviewed at latest revision, 3 unresolved discussions. drake/systems/plants/rigid_body_plant/rigid_body_plant.h, line 163 at r4 (raw file):
|
Review status: all files reviewed at latest revision, 3 unresolved discussions. drake/systems/plants/rigid_body_plant/rigid_body_plant.h, line 163 at r4 (raw file):
|
drake/systems/plants/rigid_body_plant/rigid_body_plant.h, line 163 at r4 (raw file):
|
Review status: 3 of 4 files reviewed at latest revision, 3 unresolved discussions. drake/examples/kuka_iiwa_arm/urdf/iiwa14.urdf, line 3 at r4 (raw file):
|
Reviewed 1 of 1 files at r5. Comments from Reviewable |
1 similar comment
Reviewed 1 of 1 files at r5. Comments from Reviewable |
Review status: all files reviewed at latest revision, 1 unresolved discussion. drake/systems/plants/rigid_body_plant/rigid_body_plant.h, line 163 at r4 (raw file):
|
Use cylinders instead of meshes for some links, for processing speed.
Leave meshes for links at begin and end of arm, for model accuracy.
Changed colors to better match reality. In our arm, only a part of the link
is orange, and this colors the entire link orange, but its an improvement
to see each link.
Adding mesh files in dae format and model.config for future work in Gazebo.
Replace link 5’s mesh with a cylinder.
Update collision comments
This change is