Initial commit of abb_irb2600 support

[Jmeyer1292]

This PR adds an implementation of the ABB IRB2600. I'm not entirely sure which arm length, I need to confirm.

Addressing points from @gavanderhoorn's previous critique:

1. The urdf and meshes were created with the Solidworks exporter, though I didn't do it. I'll look into the local joint transformation issues, but the over all robot works.
2. I don't know about the 'base' transformation, or even why it exists. Seems like it should be at 0,0,0. @Levi-Armstrong Do you know why this is where it is?
3. I agree on the precise model comment.

[gavanderhoorn]

Moved here from ros-industrial/abb#87.

[gavanderhoorn]

@Jmeyer1292: will you update the PR after you've confirmed the specific variant this is? In light of ros-industrial/abb#47, I feel we shouldn't accept any more support pkgs without the proper naming in place. Otherwise we'd have to update 'all' packages to use the correct names later.

[gavanderhoorn]

Regarding the base transform: it should most likely be a 'zero transform', just like all other ABB pkgs.

This pkg was probably based on the M-10iA support pkg, which has its base transform at 0.45m over Z (here).

[Levi-Armstrong]

@Jmeyer1292:

I don't know about the 'base' transformation, or even why it exists. Seems like it should be at 0,0,0.

My understanding is that the base transformation is the location of the actual base/zero location for the robot controllers, since not all robot have there origin at the mounting plate. This was added to allow for users to query between the base transformation and the end effector transformation where the position and orientation should now match between ROS and teach pendant readout.

[gavanderhoorn]

See Create a URDF for an Industrial Robot - Additional/Standard Frames for more info about this.

[Jmeyer1292]

I've updated the package to follow the new naming conventions. New move-it configs for the 20kg and 12kg variants of the 165cm model.

[gavanderhoorn]

Some comments:

• almost all launch files are missing an empty last line
• the joint_names_ yaml file in config does not have a variant specific suffix. I know the contents will be the same (as we use joint_N for all ABB robots), but this is for (future) consistency with other files in the package
• the meshes/irb2600_165 is a bit confusing (see also my comment on Define ABB specific naming rules/guidelines abb#75). I know those are the current rules, so this is not really a critique, more of an observation
• I like how you avoided duplication of the xacro macros in urdf, but apart from payload these variants are completely identical (which made your setup possible), so I would just keep only one macro and add a note to the package manifest telling users which variants are supported by that one macro
• As the variants are identical, I'd suggest keeping only a single MoveIt configuration pkg as well. A note similar to the one in the support package could be added to the manifest of the config pkg.

It could be that maximum accelerations are affected by payload, but we don't know (afaik), and the MoveIt config pkgs also don't reflect that.

[gavanderhoorn]

@Jmeyer1292: I've opened a PR against your abb_irb2600 branch with a few minor tweaks to some files I think would make sense.

The only 'issue' then is the fact that some joint transforms in the xacro macro have translations in more than a single dimension (ie: x and z fi). This is almost always an artefact of using the Solidworks urdf exporter and relying on it to derive the joint origins. In my own pkgs I've tried to avoid that by adding coordinate origins (reference geometry) and making the exporter use those.

I'll leave it to @Levi-Armstrong to decide whether that should be fixed.

[Jmeyer1292]

@Levi-Armstrong I have merged in Gijs fixes and rebased on the latest version of groovy-devel. Please advise regarding the xacro macro translations mentioned by Gijs above.

[Levi-Armstrong]

@Jmeyer1292

I would prefer that the joint position be changed, I briefly check the dimensions and some appear to be off a few millimeters. The joint per the product specification should be as below.

joint_1 (0, 0, 0.445)
joint_2 (0.15, 0, 0)
joint_3 (0, 0, 0.7)
joint_4 (0, 0, 0.115)
joint_5 (0.795, 0, 0)
joint_6 (0.085, 0, 0)
tool_0 (0, 0, 0)

[simonschmeisser]

any news on this?

[gavanderhoorn]

@Levi-Armstrong's last comment about the joint offsets still needs to be addressed and the CMakeLists.txt of the support pkg has a DEPENDENCIES argument in the roslaunch_add_file_check(..) macro that needs to be removed.

Other than that I think this could be merged (although a regen of the MoveIt pkg with the latest MSA might be also be nice).

[Jmeyer1292]

@simonschmeisser I went to the ABB website to pull down the CAD models and start over on this package, but all of the solidworks assemblies are gone. Will ask this week about my CAD options.

[Jmeyer1292]

Alrighty. Two years later, here I am. I have updated the xacro file to match the spec sheet and our conventions. I re-exported all of the cad models and regenerated the convex hulls using blender.

I removed the dependency stuff @gavanderhoorn mentioned. And I entirely dropped the moveit config package as I'd rather not maintain it.

[AustinDeric]

depreciated

[Jmeyer1292]

It hasn't lost that much value to wear and tear.