Skip to content

Latest commit

 

History

History
84 lines (72 loc) · 4.19 KB

README.md

File metadata and controls

84 lines (72 loc) · 4.19 KB
Raw

Cartesian Controller Utilities

This package turns the 3DConnexion (spacenav_node) from the joystick_drivers into a force and motion control input device for teleoperation with the cartesian_controllers:

  • Force control: Re-publish geometry_msgs/Twist messages as geometry_msgs/WrenchStamped messages to intuitively control the target_wrench.
  • Motion control: Turn geometry_msgs/Twist messages into a geometry_msgs/PoseStamped to smoothly steer the robot via its target_frame.
  • Button commands: Define custom commands for the space mouse buttons with ROS command line instructions. You can trigger events, such as calling ROS services and publish to topics.

Install

We need some additional system dependencies

sudo apt install libspnav-dev spacenavd screen

and these Python dependencies for the scripts:

pip3 install numpy numpy-quaternion

We also need a custom version of the joystick_drivers (one that correctly reports the button events). Until pending PRs are merged upstream there, you'll find the package here. Navigate into your src folder in your workspace,

git clone -b ros2 https://github.com/stefanscherzinger/joystick_drivers.git

and build the workspace as usual.

Force control

Start the simulation and the force controller as described here. In a sourced terminal, call

ros2 launch cartesian_controller_utilities spacenav_to_wrench.launch.py

You can now steer the robot's end-effector with the space mouse. There's a parameter to change the coordinate frame for control: In rqt open the Dynamic Reconfigure plugin under Configuration, navigate to my_cartesian_force_controller and untick the box for hand_frame_control. You now steer the robot in the controller's robot_base_link coordinates.

Motion control

Start the simulation and the compliance controller as described here. In a sourced terminal, call

ros2 launch cartesian_controller_utilities spacenav_to_pose.launch.py

You can now steer the robot's target_frame with the space mouse. The publisher makes sure that we always start from the current end-effector pose. We use the cartesian_compliance_controller in this example, because it's suitable for teleoperation that involves contact with the environment and is thus more versatile than the cartesian_motion_controller. You can take this example and adjust the setting for your own robot manipulator.

Button commands

The space mouse's buttons are extremely handy for triggering custom events during teleoperation, such as opening and closing grippers. You can program both buttons with everything that's a valid (ROS) command on the command line. In this example, we record the spacenav/twist topic into a rosbag and use the two buttons for starting and stoping, respectively. In a source terminal, call

ros2 launch cartesian_controller_utilities spacenav_buttons.launch.py

Now press the space mouse's buttons. You should see output in the terminal, similar to this

[buttons.py-2] start recording
[buttons.py-2] stop recording

Per default, the rosbags will end up in the terminal's current folder. A practical use case could be recording the target_frame during teleoperation for motion replay with the cartesian_motion_controller.

Disable GUI control in Gazebo

When working with the 3DConnexion as a force control input device, you probably don't want to control gazebo's camera at the same time. Remember your kernel drivers still treat your force control thing as a space mouse, which is used by gazebo by default. To switch this behavior off add the following lines to the ~/.gazebo/gui.ini file:

[spacenav]
deadband_x = 0.1
deadband_y = 0.1
deadband_z = 0.1
deadband_rx = 0.1
deadband_ry = 0.1
deadband_rz = 0.1
topic=~/spacenav/remapped_joy_topic_to_something_not_used

More information can be found here.