Update to latest tesseract

[marrts]

Updating to the latest version of Tesseract, including the latest tesseract_ros2 changes.

• Of particular note is the new plugin-based taskflow setup

Currently WIP while waiting for tesseract-robotics/tesseract_planning#314 to be resolved.

[marrts]

Updated to latest commit of all tesseract branches.

Currently a bug with custom task constant tcp time param, but it does run.

[marrts]

This is now fully working, but failing CI because it is using an older tesseract docker image to build off of.

[marip8]

Are these actually used in this file? Also, do are these already defined somewhere in a tesseract header? If so and if we need these values, let's not redefine them

[marrts]

These directly correspond to the names given in the task_composer_plugins.yaml file. They are not hardcoded any where in Tesseract, but I did copy the names from the example in Tesseract. Perhaps that should be made more clear.

[marip8]

I'm trying to build this now, and it seems like this PR actually targets the master branch of tesseract_planning not 0.17.0 and not 0.16.3 (from the dependencies.repos). Can you update the name of the PR and the .repos file to the correct version?

[marrts]

I'm trying to build this now, and it seems like this PR actually targets the master branch of tesseract_planning not 0.17.0 and not 0.16.3 (from the dependencies.repos). Can you update the name of the PR and the .repos file to the correct version?

Updated

[marrts]

I'm suddenly getting IK solver failures, I need to figure out what has happened.

[marrts]

I'm suddenly getting IK solver failures, I need to figure out what has happened.

It seems like something changed to somehow break the IKfast implementation. If I use our generated IKFast the motion planner crashes, but it works just fine with KDL.

I tested the code before this PR and confirmed IKFast was working. It would generate 4 solutions per IK and when I switched to KDL it would only generate 1 (it's also obvious when using KDL because the terminal populates with a bunch of extra text).

I also duplicated the IKFast crashing on another machine that was working fine with the code before this PR.

Tesseract hasn't had any changes to the IKFast code in the tesseract_kinematics package and I confirmed that it was finding and loading our generated motoman ikfast solver by adding debugging print statements in tesseract where plugins are loaded.

When I attach a debugger to the motion planner it crashes at this line because I guess some values are invalid: https://github.com/marrts/scan_n_plan_workshop/blob/789110816e1805b373f8368146d6344077714451/snp_motion_planning/src/plugins/kinematics/motoman_hc10_ikfast_solver.hpp#L7512

I haven't figured out why this is happening yet.

[marrts]

@marip8 This is passing CI for Foxy now, so it should be ready to merge and we can make issues about getting humble and rolling passing CI.

Although, I'm not sure why clang-format is failing. I don't see anything in the CI output and nothing changes locally for me when I run clang format.

[rjoomen]

Could the IKFast crashing have to do with the LAPACK dependencies of qpOASES, which was recently added?

[marrts]

Could the IKFast crashing have to do with the LAPACK dependencies of qpOASES, which was tesseract-robotics/trajopt#327?

I'm not familiar with those libraries and how they might impact IKFast. Is there a specific reason why you think these might be causing a problem?

Also, I know that you've been working on a PR related to IKFast in Tesseract, did you experience any similar issues, or is that working smoothly?

[rjoomen]

I did not experience similar issues, but I checked, and my IKFast solution does not use any of the LAPACK routines, and neither do the solvers used for the unit tests in tesseract_kinematics (abb_ird2400 and iiwa7).

It just occurred to me this might be the problem as it could be a dependency issue, as nothing did change to your code.