Test out iiwa_ros2 for victor

[PeterMitrano]

https://github.com/ICube-Robotics/iiwa_ros2 Steps:

• [x] Try out the iiwa_ros2 repo instructions
  • [x] Get one arm working without the gripper
    • [x] start by forking the iiwa_ros2 repo into the armlab org
    • [x] make an updated URDF with both arms. Probably start with the new urdf because theirs is very pretty, and just reference our victor.urdf to make a new victor.urdf. For now we can ignore the gripper in the urdf.
    • [x] In IiwaFRIHardwareInterface, create two robotClient_ objects, one for left and one for right, and update the code to handle that. We can just assume a fixed ordering (left then right) in the vectors.
    • [x] include all 14 joints (this will require renaming them!) here and here
    • [ ] Upgrade arm_robots package files to ros2
    • [ ] Get the victor basic demo script to run

For the robotiq 3f grippers

• [ ] Get a cable that can connect from the X651 port on the kuka to kukacontrol (ethernet? or usb?)
• [ ] Make a working ros2 driver for the robotiq 3 by combining ros2_robotiq_gripper and robotiq_3f_gripper_control
• [ ] add the robotiq grippers in the URDF

[PeterMitrano]

So it turns out that using the gripper is actually the tricky part here. Using the iiwa_ros2 java file is really simple, I got it working in the sense that the FRI interface is started and waiting for commands. So some next steps are:

• [ ] Finish setting up the C++ FRI ROS2 side of things on the kukacontrol machine
• [x] Calibrate the gripper since the arm still needs to know the mass of the robotiq gripper even though it's not going to be controlling it directly anymore
• [ ] Get the robotiq working. This means getting some code on the kukacontrol machine that works and making/using the right cables to connect the X651 port to the kuka control machine + power

[PeterMitrano]

Yating and I started on setting up the FRI stuff. We added the gripper as a tool in the sunrise project and ran load calibration. We had to set real values for stiffness and damping in the iiwa_ros2.java file which is weird, because we thought those values should be overwritten by control via FRI, but that doesn't seem to be happening. Position control works, as does velocity control. We mostly tested via ros2 topic publish or the rqt_joint_trajectory_controller. Planning and execution via MoveIt also works for the left arm in isolation.

However, the bad news is joint2 on the left arm is now making a crunchy bad noise. Not sure what we did to it, but it's not throwing any specific errors. We think this happened when using rqt_joint_trajectory_controller so probably avoid using this.

[PeterMitrano]

I think a next step is to try to modify the controller to control both arms. I added some steps above (probably there are more)

[bsaund]

However, the bad news is joint2 on the left arm is now making a crunchy bad noise.

FYI this happened before and we sent the robot back to Kuka for servicing. We had a loaner white kuka iiwa for awhile. I don't remember which arm

[PeterMitrano]

abandoning this in favor of just upgrading the existing stack to ROS 2 directly