I added transform and message to publish camera pose to /mavros/vision_pose/pose topic to use with mavros package and a multirotor with Arducopter for autonomous indoor flight. I added also a launch file detection_rpi.launch that use gscam to get video streamed with gstreamer.
Moreover create_board, useful to create a board as image file, has been modified to write a layout.yaml file (to stdout) to use with the package, see layout-my.yaml as example.
This node detect the ARUCO board that is on the target. It then publishes the corresponding pose in the /vision/pose topic.
Clone and build OpenCV from source with the extra modules
$ cd <opencv_build_directory>
$ cmake -DOPENCV_EXTRA_MODULES_PATH=<opencv_contrib>/modules <opencv_source_directory>
$ make -j5
To use aruco with ROS, we need to use opencv 3.2. Since that we need cv_bridge, we have to rebuild vision_opencv from source. Otherwise cv_bridge is built with opencv2. link
You can clone vision_opencv in your catkin_ws/scr and after build with:
$ cd ~/catkin_ws
$ catkin_make -DOpenCV_DIR=/path-to-build-opencv-3.2
When catkin_make has finish, you can check if the cv_bridge is using the right version of OpenCV with the following commands:
$ cd ~/catkin_ws/devel/lib
$ ldd libcv_bridge.so | grep opencv
libopencv_core.so.3.2 => /home/jokla/Software/opencv-3.2.0/build/lib/libopencv_core.so.3.2 (0x00007f7acb240000)
libopencv_imgcodecs.so.3.2 => /home/jokla/Software/opencv-3.2.0/build/lib/libopencv_imgcodecs.so.3.2 (0x00007f7acaffe000)
libopencv_imgproc.so.3.2 => /home/jokla/Software/opencv-3.2.0/build/lib/libopencv_imgproc.so.3.2 (0x00007f7ac97ca000)
Now you can clone aruco_gridboard in your catkin_ws and build with catkin_make
$ cd ~/catkin_ws/src
$ git clone https://github.com/jokla/aruco_gridboard.git
$ cd ..
$ catkin_make
$ source devel/setup.bash
Add the new OpenCV to the path:
$ export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/jokla/Software/opencv-3.2.0/build/lib/
Launch the detection node:
roslaunch aruco_gridboard detection.launch
Here you can see an example of the launch file:
<launch>
<!-- Launch the detection node -->
<node pkg="aruco_gridboard" type="aruco_gridboard" name="aruco_gridboard" output="screen" >
<param name="board_path" value="$(find aruco_gridboard)/data/layout.yaml" />
<param name="detector_param_path" value="$(find aruco_gridboard)/data/detector_params.yml" />
<param name="debug_display" value="True" />
<remap from="/aruco_gridboard/camera_info" to="/camera/camera_info"/>
<remap from="/aruco_gridboard/image_raw" to="/camera/image"/>
</node>
</launch>
The node is subscribing to the image topic /camera/image and the topic /camera/camera_info containing the camera parameters. Aruco will try to detect the board described by the yaml file (you can set in with the parameter board_path and it will publish the board pose on the topic /vision/pose and the detection status on the topic /vision/status.