apriltag_ros

apriltag_ros is a Robot Operating System (ROS) wrapper of the AprilTag 3 visual fiducial detector. For details and tutorials, please see the ROS wiki.

apriltag_ros depends on the latest release of the AprilTag library. Clone it into your catkin workspace before building.

Authors: Danylo Malyuta, Wolfgang Merkt

Maintainers: Danylo Malyuta (Autonomous Control Laboratory, University of Washington), Wolfgang Merkt

Quickstart

Starting with a working ROS installation (Kinetic and Melodic are supported):

export ROS_DISTRO=melodic               # Set this to your distro, e.g. kinetic or melodic
source /opt/ros/$ROS_DISTRO/setup.bash  # Source your ROS distro 
mkdir -p ~/catkin_ws/src                # Make a new workspace 
cd ~/catkin_ws/src                      # Navigate to the source space
git clone https://github.com/AprilRobotics/apriltag.git      # Clone Apriltag library
git clone https://github.com/AprilRobotics/apriltag_ros.git  # Clone Apriltag ROS wrapper
cd ~/catkin_ws                          # Navigate to the workspace
rosdep install --from-paths src --ignore-src -r -y  # Install any missing packages
catkin_make_isolated
source devel_isolated/setup.bash

See the ROS wiki for details and tutorials.

Tag Size Definition

For a correct depth estimation (and hence the correct full pose) it is necessary to specify the tag size in config/tags.yaml correctly. In the Wiki for the AprilTag Library the correct interpretation of the term "tag size" is explained. The size is defined by the length of the black/white border between the complete black and white rectangle of any tag type. Note that for apriltag3 marker families this does not in fact correspond to the outside of the marker.

Below is a visualization of the tag size (red arrow) to be specified for the most common tag classes:

Contributing

Pull requests are welcome! Especially for the following areas:

Publishing of the AprilTag 3 algorithm intermediate images over a ROS image topic (which AprilTag 3 already generates when tag_debug==1)
Conversion of the bundle calibration script from MATLAB to Python
Extend calibration to support calibrating tags that cannot appear simultaneously with the master tag, but do appear simultaneously with other tags which themselves or via a similar relationship appear with the master tag (e.g. a bundle with the geometry of a cube - if the master is on one face, tags on the opposite face cannot currently be calibrated). This is basically "transform chaining" and potentially allows calibration of bundles with arbitrary geometry as long as a transform chain exists from any tag to the master tag
Supporting multiple tag family detection (currently all tags have to be of the same family). This means calling the detector once for each family. Because the core AprilTag 2 algorithm is the performance bottleneck, detection of n tag families will possibly decrease performance by 1/n (tbd if this still holds for v3)

Changelog

In March 2019, the code was upgraded to AprilTag 3 and as thus the options refine_pose, refine_decode, and black_border were removed.

Copyright

The source code in apriltag_ros/ is original code that is the ROS wrapper itself, see the LICENSE. It is inspired by apriltags_ros and provides a superset of its functionalities.

If you use this code, please kindly inform Danylo Malyuta (to maintain a list here of research works that have benefited from the code) and cite:

D. Malyuta, C. Brommer, D. Hentzen, T. Stastny, R. Siegwart, and R. Brockers, “Long-duration fully autonomous operation of rotorcraft unmanned aerial systems for remote-sensing data acquisition,” Journal of Field Robotics, p. arXiv:1908.06381, Aug. 2019.
C. Brommer, D. Malyuta, D. Hentzen, and R. Brockers, “Long-duration autonomy for small rotorcraft UAS including recharging,” in IEEE/RSJ International Conference on Intelligent Robots and Systems, IEEE, p. arXiv:1810.05683, oct 2018.
J. Wang and E. Olson, "AprilTag 2: Efficient and robust fiducial detection," in ''Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)'', October 2016.