ROS Apriltag detector package

This package has the base class header files for apriltag detector plugins for the UMich and the MIT detector.

It also has launch files for both of these detectors, so by installing this package you can use both of these detectors interchangeably. Note that UMich detector is more sensitive and faster, but it does not work when tags are encroached by black markers or have double-wide black borders (Kalibr boards!). The MIT detector must be used in the latter situation.

For more documentation on how to install this package refer to the documentation of the apriltag_detector repository.

NOTE: Because running an apriltag detector is a heavy weight operation, the detector is very frugal about subscribing to image topics. Unless you bring up e.g. an image viewer to connect to the debug images, the detector will just sit there and do nothing.

Launch files

detect.launch.py

This convenience launch file launches both a detector and a draw node for easy viewing of the tags.

Arguments:

• black_border_width: (only for MIT). Width (in bits) of the outer black border of the apriltags. Note that Kalibr board tags often have border width of 2 bits. Default: 1.
• blur: (only for UMich). Gaussian blur sigma (in pixels). Default: 0 (no blur).
• camera: name of the camera, e.g. /camera_0. Default: camera.
• image: name of the image underneath the camera node, e.g. image_raw. The node will then subscribe to images /camera_name/image_raw. Default: image_raw.
• image_transport: the transport to use, e.g. compressed, ffmpeg etc. Default: raw.
• max_allowed_hamming_distance: (only for UMich) when more than this number of bits are different from a proper code, disregard the tag. Default: 0 (perfect match required). Default: raw.
• tags: The topic name under which to publish the tags. Default: tags.
• type: What type of detector to use. Valid are mit, umich. Default: umich.

Example usage

Start detector and drawing node:

ros2 launch apriltag_detector.launch.py camera:=/cam_sync/cam_0

Now you have to start a camera server that publishes images under /cam_sync/cam_0/image_raw, or play such data from a bag.

For the detector to do anything you also have to subscribe to one of its output topics, for instance with rqt_image_view:

ros2 run rqt_image_view rqt_image_view /cam_sync/cam_0/image_tags

Alternatively you can look at the decode tag rate:

ros2 topic hz /cam_sync/cam_0/tags

License

This software is issued under the Apache License Version 2.0.

Changelog for package apriltag_detector

2.1.0 (2024-05-11)

• switched to plugins, added MIT detector, split off apriltag_draw
• Contributors: Bernd Pfrommer