SARAFun force learning

This code is developed in the context of the SARAFun project

Content

• ibd_force_teaching: main nodes enabling the extraction of the insertion by deformation characteristics.
• ibd_test_data: fake data to be used for testing the node
• force_teaching_msgs: specific ROS communication definition.

Use

Warning: The code is still under development.

Launch the force teaching module

rosrun ibd_force_teaching ibd_force_teaching
# to launch an action interface
rosrun actionlib axclient.py /learn

To adjust the module to a larger application, the minimum parameters to set are the followings:

rosrun ibd_force_teaching ibd_force_teaching _force_frame:=/[tf_force_frame] _receptacle_object_frame:=/[tf_object_frame] wrench:=[wrench_topic_name]

with:

• force_frame : the tf frame related to the force sensor
• receptacle_object_frame : the tf frame related to the object into which the deformable object will be inserted.
• wrench: name of the topic containing the WrenchStamped measured by the force sensor.

Launch with fake data

This will launch the data stored in ibd_test_data.

roslaunch ibd_test_data pub_fake_data.launch
rosrun ibd_force_teaching ibd_force_teaching
# to launch an action interface
rosrun ibd_test_data bridge_wrench_action _ibd_learn_remap:=/learn

The last script, bridge_wrench_action detect when the whole wrench data has been published. At that moment, it launches the action of learning. The data reproduction (pub_fake_data.launch) once published everything, wait for one second, and thn resume the publication from the beginning. So that the learning is started at the beginning of the data.

Looking at the data published (focus on the third row, the force measured, where 6, 7, 8 respectively refers to x y and z dimension):

The two magnitudes looked at (and returned by the action) should be around 6N (max_force_deformation in the action, and orientation in the image) and 14N (max_force_snap in the action result and insertion on the figure).

The illustrative files is generated using:

python ar_sarafun_ibd.py --c config/trial_deformation.yaml