ft-sensor-calibration

ROS 2 calibration procedures for Torque-Force sensor in Agimus project

Install

vcs import --shallow --recursive src < src/panda_ft_sensor_calibration/dependencies.repos

rosdep update --rosdistro $ROS_DISTRO
rosdep install -y -i --from-paths src --rosdistro $ROS_DISTRO
# parameter --symlink-install is optional
colcon build --symlink-install

Static calibration

First record data

ros2 launch panda_ft_sensor_calibration jtc_follow.launch.py robot_ip:=<ip of the robot> ft_sensor_ip:=<ip of the ft sensor> output_rosbag_path:=<path to store rosbag>

Then preprocess it

ros2 run rosbag2pandas rosbag2pandas <path to a rosbag from previous step> <path to store parsed data> -s mcap --format parquet

After all those steps are done move to a notebook mass_calibration.ipynb and follow steps there.

Dynamic calibration

First manually record the motion

ros2 launch panda_ft_sensor_calibration record_motion.launch.launch.py robot_ip:=<ip of the robot> ft_sensor_ip:=<ip of the ft sensor> output_rosbag_path:=<path to store a rosbag with manual motion>

Then rerecord that motion without external forces from a human

ros2 launch panda_ft_sensor_calibration playback_with_sensor_recording.launch.launch.launch.py robot_ip:=<ip of the robot> ft_sensor_ip:=<ip of the ft sensor> output_rosbag_path:=<path to store rosbag with replayed> input_rosbag_path:=<path to a rosbag from previous step>

Then preprocess the data

ros2 run rosbag2pandas rosbag2pandas <path to store rosbag> <path to rosbag with replayed motion> -s mcap --format parquet