This system will use a usb webcam to detect and locate Aruco markers. The pose of each marker will be returned on a topic in the ‘/transformed’ namespace, followed by the marker that topic is tracking. E.g. The pose of the Aruco marker with code 1 is published on the ‘/transformed/marker1’ topic. The pose of each maker is published as a Pose message type (geometry_msgs/msg/Pose) which contains the position and orientation of a given marker relative to marker 0, shown below. The position is in cartesian (X,Y,Z) coordinates and the orientation is given in quaternions in the order x,y,z,w. Each element can be retrieved using the following syntax: Pose.position.X or Pose.orientation.x All the distances are measured in metres from the centre of each marker.
The 'joint_test' file is not needed and can be ignored.
- navigate to the computers root directory
- create a ROS2 workspace and navigate into the /src folder
- clone this repo along with the 'aruco_ros' and 'usb_cam' repos
aruco_ros: https://github.com/pal-robotics/aruco_ros usb_cam: https://github.com/ros-drivers/usb_cam
- navigate back to the workspace, install dependancies, and build
Full instructions for this can be found on the ROS documentaion website as well as in the repositories mentioned above.
Once all the code had been correctly installed, the camera will need calibrating. To do this the camera parameters YAML file must be edited. First enter the resolution that you are using for the camera into the image height and width fields, and set the distortion model to plumb bob (this is the case for most cameras). Next, a tool such as the MATLAB Camera Calibrator App https://uk.mathworks.com/help/vision/ref/cameracalibrator-app.html can be used to obtain the internal matrix K, and the radial and tangential distortion coefficients.
The camera matrix and rectification matrix should both be set to the internal matrix K, and the distortion coefficients should be a vector containing the 2 radial, then the 2 tangential distortion coefficients with a final 0. The projection matrix will should equal the internal matrix K followed by a final column of zeros.
The usb_cam package comes with a default set of camera parameters, but you can use any file for this by specifying the file directory in the 'camera_info_url' field in the launch_pkg/launch/vison_launch.py file.
To run, use the terminal to go to, then source, the correct workspace.
Source install/local_setup.bash
Then run the ‘vision_system_launch’ script.
Ros2 launch launch_pkg vision_launch.py
This will launch the ‘USB cam’ node, 4 ‘Aruco double’ nodes, and the ‘pose transformer’ node. This will allow the system to track 6 blocks as well as the 2 larger markers ( codes 0, 1) used to find the position of both tables. The pose for each marker is published every time the system receives a frame from the camera. Current camera frame rate is 10 Hz, so this also around the rate at which marker poses are published.