 Please ask about problems and questions regarding this tutorial on answers.ros.org. Don't forget to include in your question the link to this page, the versions of your OS & ROS, and also add appropriate tags. |
Configure cob_sensor_fusion
Description: This tutorial teaches the configuration of the parameters for cob_sensor_fusion.Tutorial Level: BEGINNER
Next Tutorial: How to train a new object
These tutorials require packages that are not open to the public. If you want to run the packages, please contact richard.bormann <at> ipa.fraunhofer.de . |
Contents
How to run and configure cob_sensor_fusion, using stereo cameras
You can adjust the parameters for cob_sensor_fusion to calculate the depth image from a stereo camera system as follows.
Startup
At first you should start a robot instance. When using a real Robot have a look at the Care-O-bot Manual and the cob_bringup package to bringup a real Care-O-bot. Otherwise, when using the Gazebo simulation have a look at the cob_bringup_sim package to bringup Care-O-bot in simulation.
Then run cob_sensor_fusion on pc2 using the stereo camera pair
roslaunch cob_sensor_fusion sensor_fusion_stereo.launch
Dynamic Reconfigure
The dynamic reconfigure interface adjusts the parameters at runtime. To start dynamic reconfigure type
rosrun dynamic_reconfigure reconfigure_gui
Then you can select the ROS node that offers the dynamic reconfigure parameter. Select topic /sensor_fusion/stereo/sensor_fusion and you should see a window like shown below:
Playing with the stereo parameters
Try to improve the parameters by using the sliders or text fields. When selecting 'advertise_depth_image' you can display the resulting depth image with:
rosrun image_view image_view image:=/sensor_fusion/stereo/disparity
Adjust the parameters until you have reached a suitable trade-off between performance in computation time (fps) and range image quality.
The following image shows the resulting range data when using the parameter setting as listed above:
NOTE: When you cannot see any reasonable range data, your stereo cameras might be not configured properly and you might consider recalibrating them using e.g. cob_calibration package.
