Running demo Tiago deburring

Hardware

• Tiago robot,
• part P72,
• driller,
• laptop tatesan (login jmirabel, passwd)

Steps to run the demo

The demonstration can be run in simulation in Gazebo. Specific steps for that are prefixed by simulation:.

1. plug a computer with the docker image gitlab.laas.fr:4567/rob4fam/docker/tiago_deburring:1_vnc_robot to Tiago with ethernet cable and select wired connection "Tiago".

2. start a docker container and a VNC client and run the commands of the next steps in the docket container

docker run -it --rm --name hpp --cap-add NET_ADMIN --net=host gitlab.laas.fr:4567/rob4fam/docker/tiago_deburring:1_vnc_robot

Make sure that file /etc/hosts contains the following line:

10.68.0.1       tiago-48c

3. simulation: in terminal 0

  roslaunch agimus_demos tiago_deburring_simulation.launch

4. in terminal 1 launch hppcorbaserver
5. in terminal 2

  cd agimus-demos/tiago/deburring
  python2 -i travelling_salesman.py

6. simulation: in terminal 3

  cd agimus-demos/tiago/deburring
  python stop-controllers.py

6. in terminal 3

   ssh pal@tiago-wired
   stop-controllers

and check that all relevant controllers have been successfully stopped.

7. simulation: in terminal 3

  roslaunch agimus_demos tiago_deburring_demo.launch estimation:=false simulation:=true

7. in terminal 3

  roslaunch agimus_demos tiago_deburring_demo.launch estimation:=false

8. in terminal 4

  ssh pal@tiago-wired
  roslaunch agimus_demos tiago_deburring_estimation.launch

9. in terminal 5

  cd agimus-demos/tiago/deburring
  roslaunch ./rviz.launch

10. simulation: in terminal 6

  python initial_pose.py

interrupt the script when the robot is correctly localized in rviz

10. check that the robot is well localized. If not,

• click in rviz on button 2D pose estimate and select an approximate pose with the mouse. Move the robot around its current position until the localization candidates converge to a reasonable value, or
• rosservice call /global_localization

11. in terminal 6

  rosrun agimus rqt_path_execution

12. in terminal 7

  gepetto-gui

13. in terminal 2

bpath = compute_base_path_to_cluster_init(0)
ps.hppcorba.problem.addPath(bpath) # Returns the path ID in HPP

You can visualize the trajectory after refreshing gepetto-gui. If you are satisfied, you can execute the trajectory using the rqt_path_execution window. The robot starts closing the gripper. Don't forget to put the driller in the hand.

You may have to recompute the clusters if HPP fails to reach the handles. After having recomputed the clusters, you must move the base again, as above (step 12).

clusters = recompute_clusters()

You can now compute a motion of the arm to reach the holes.

ncluster, apath = compute_path_for_cluster(0)
# ncluster contains the new cluster (reached holes, with the updated configs)
# apath is the wholebody trajectory
apath_id = ps.hppcorba.problem.addPath(apath)
ps.optimizePath(apath_id)
# Visualize the last path in gepetto-gui

Instead of step 9, you can run the script

python initial_pose.py

That will send the initial pose of the robot to the navigation stack.