Dev/ros api/calibration

[Ruthrash]

Probable stable features in this pull request

• [x] bug free build of realtime computer docker environment
• [x] bug free build of workstation computer docker environment
• [x] bug free usage of frankapy using both the realtime and workstation docker environments tested with frankapy/scripts/reset_arm.py script
• [x] error free documentation README.md to build and use the docker environments
• [x] ROS based script to collect pose data for calibration
• [x] non ROS based script to collect pose data for calibration

[kouroshD]

On the real_time machine side, still there is problem with ROS_MASTER_URI, ROS_HOSTNAME; and ROS_IP. roscore runs on the workstation machine. have a look on this link for correctly setting the those variables.

[kouroshD]

Also, related to the reseting the robot arm as the test case, there is the following problem. since, we do not have franka gripper, reset_arm does not work and it will block at fa=FrankaArm(). To check, till this point was OK, I ran the following script:

root@sf3202msi-Z490-A-PRO-Aegis-RS:~/git/frankapy# python3
Python 3.8.10 (default, Nov 14 2022, 12:59:47) 
[GCC 9.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from frankapy import FrankaArm
WARNING:root:autolab_core not installed as catkin package, RigidTransform ros methods will be unavailable
>>> fa = FrankaArm( with_gripper=False)
>>> fa.reset_pose()

[Ruthrash]

On the real_time machine side, still there is problem with ROS_MASTER_URI, ROS_HOSTNAME; and ROS_IP. roscore runs on the workstation machine. have a look on this link for correctly setting the those variables.

strange that this error pops up as startup scripts from frankapy set these environment variables in the realtime computer. see below

• https://github.com/iamlab-cmu/frankapy/blob/4e05076fa1022ca04d4b9564b67f6a028f60645c/bash_scripts/start_franka_ros_interface_on_control_pc.sh#L24
• https://github.com/iamlab-cmu/franka-interface/blob/master/bash_scripts/set_rosmaster.sh

[kouroshD]

strange that this error pops up as startup scripts from frankapy set these environment variables in the realtime computer. see below before setting them manually, the ROS_MASTER_URI was set to localhost, and ROS_HOSTNAME and ROS_IP were empty.

btw, here https://github.com/iamlab-cmu/franka-interface/blob/780a6c81983c83a93d6ed72b413db5b325442343/bash_scripts/set_rosmaster.sh#L8 the ROS_IP is set to the realtime machine ip whereas I set it in our case to where roscore was running, i.e., the same as ros master uri. Looking at http://wiki.ros.org/ROS/EnvironmentVariables#ROS_IP.2FROS_HOSTNAME, it explains why it was working, ros_hostname takes precedence over ros_ip when both are set. So, we should update this later in our real_time machine by setting ros_ip as real_time machine ip.

[kouroshD]

On the real_time machine side, still there is problem with ROS_MASTER_URI, ROS_HOSTNAME; and ROS_IP. roscore runs on the workstation machine. have a look on this link for correctly setting the those variables.

strange that this error pops up as startup scripts from frankapy set these environment variables in the realtime computer. see below

• https://github.com/iamlab-cmu/frankapy/blob/4e05076fa1022ca04d4b9564b67f6a028f60645c/bash_scripts/start_franka_ros_interface_on_control_pc.sh#L24
• https://github.com/iamlab-cmu/franka-interface/blob/master/bash_scripts/set_rosmaster.sh

coming back again to this point, since we are sourcing the terminals that are opened by this script, it does not affect other terminals. So, in my case, when I opened a new terminal in realtime machine, I could not run communicate with the workstation machine. But we are not normally opening new terminals in realtime machine. Proabably you are right, it might be the case only for the with_gripper=False should fix the problem, without the need to set the env variables. to be checked!

[n-yoshikawa]

I confirmed the docker scripts are compatible with docker-compose v2.

[kouroshD]

wha is the difference between https://github.com/pairlab/franka_arm_infra/blob/dev/rosAPI/calibration/calibration/README.md and https://github.com/pairlab/franka_arm_infra/blob/dev/rosAPI/calibration/calibration/docs/USAGE.md?

[Ruthrash]

wha is the difference between https://github.com/pairlab/franka_arm_infra/blob/dev/rosAPI/calibration/calibration/README.md and https://github.com/pairlab/franka_arm_infra/blob/dev/rosAPI/calibration/calibration/docs/USAGE.md?

Usage will have instructions specific to using our calibration tool. As for README, I havent' completed it yet, but in there I am thinking of providing a light theoretical background to the problem + best practices (aimed at people who are completely new to doing the hand-eye calibration) the particular file you are referring to is a clone and it has been removed now

[kouroshD]

documentation missing for calibration:

• first give a very brief and general overview of the calibration process;

  • like printing april tags;
  • run tf related custom code by user if not using a standard franka robot or if using another robot;
  • running scripts for calibration;
  • and then what the user should do with the returned calibration matrix. Also I did not see where it was written where to put the april tag boards in our experiments, we did not use your april tag pdf

• I think in general it would be nice to have images coming from camera visualized so that the user knows if the board is completely visible or not to the camera;

• we printed the april board US letter, and measured the sizes. When printed out, we had following values:

      markerLength = 0.0252
      markerSeparation = 0.0049 

  Which are different from the one written in : https://github.com/pairlab/franka_arm_infra/blob/9d56e14a1c41e3c8fe878c87dcff83cbce7b05bd/calibration/ROSCameraRobotCalibration.py#L36-L37

[n-yoshikawa]

The gripper problem mentioned in https://github.com/pairlab/franka_arm_infra/pull/18#issuecomment-1469011432 should be solved by changing the line 32 of bash_scripts/start_control_pc.sh from with_gripper=1 to with_gripper=0. This is the solution used in our lab computer.

[Ruthrash]

documentation missing for calibration:

• first give a very brief and general overview of the calibration process;

  • like printing april tags;
  • run tf related custom code by user if not using a standard franka robot or if using another robot;
  • running scripts for calibration;
  • and then what the user should do with the returned calibration matrix. Also I did not see where it was written where to put the april tag boards in our experiments, we did not use your april tag pdf
• I think in general it would be nice to have images coming from camera visualized so that the user knows if the board is completely visible or not to the camera;
• we printed the april board US letter, and measured the sizes. When printed out, we had following values:

        markerLength = 0.0252
        markerSeparation = 0.0049 

Which are different from the one written in :

https://github.com/pairlab/franka_arm_infra/blob/9d56e14a1c41e3c8fe878c87dcff83cbce7b05bd/calibration/ROSCameraRobotCalibration.py#L36-L37

• added suggestions to READMEs.
• allowing user to pass arucoboard parameters as arguments to constructors.

[n-yoshikawa]

• [x] We have tried with and we had the following error, because the file of the robot waypoints are missing as thrown below.

  python3 ROSCameraRobotCalibration.py --move_robot_automatically True
  FileNotFoundError: [Errno 2] No such file or directory: 'data/file_moves.txt'

  there is a file https://github.com/pairlab/franka_arm_infra/blob/dev/rosAPI/calibration/calibration/data/file.txt that is apparently empty and not used. Maybe, you want to remove it.

• [x] In order to run the code script as mentioned above, we had to change the script in the following line: https://github.com/pairlab/franka_arm_infra/blob/4f57a072045c8734a8a582030f7b13696f13d9b3/calibration/ROSCameraRobotCalibration.py#L36 adding the option : with_gripper=False

• [ ]In the workstation docker env, we are missing the franka_arm_infra repo. However, we had the /git/calibration folder, but it is not a github repo. so we had to clone it manually. the calibration folder inside the franka_arm_infra is mounted/linked to the /git/calibration folder inside the docker so changes made in either sides(inside/outside docker) will remain consistent in both the sides

• [ ] as suggested in this comment https://github.com/pairlab/franka_arm_infra/pull/18#issuecomment-1470257162 , we tried passing the -- with_gripper 0 to the bash script in order to avoid enabling the robot gripper when starting libFranka/Franka-Interface. but now within the docker env it is not working probably due to the change the bash script files. But, this option works in our old environment without using docker and with the non-modified bash script. So, if this option is correctly working, then we can run reset_arm.py script without any changes.

[kouroshD]

Hi @Ruthrash , I had made some change requests before your recent changes but I forgot to submit my change request and that's why now they are suggested on an outdated file. Please kindly go through them; I think they are helpful. Thanks

[Ruthrash]

Thanks a lot for the reviews @n-yoshikawa and @kouroshD