This project is a ROS wrapper for the EPFL's chilitags library.

The chilitags C++ library is a robust and lightweight library for detection of fiducial markers.

Main features:

• recognizes up to 1024 markers simultaneously in a scene,
• has a focus on robustness to bad light conditions,
• provides (filtered) 6D localization of the markers,
• supports objects with multiple markers (with increased robustness),
• OpenCV (>=2.3) is the only dependency.

This ROS node wraps the chilitags library to use the standard ROS mechanisms: images and camera calibration are read from a standard ROS camera, and 6D position of markers are published as TF transforms.

Cloning this repo inside a catkin workspace will download the chilitags library, and compile it together with the ROS nodes:

cd ~/catkin_ws/src # Or wherever you catkin workspace is
git clone https://github.com/chili-epfl/ros_markers.git
catkin_make --pkg chilitags_catkin ros_markers

An example usage is:

$ roslaunch ros_markers detect.launch image_topic:=v4l/camera/image_raw camera_frame_id:=v4l_frame

(note that if you want to launch the node directly, ie, without roslaunch, this example become: ./detect image:=v4l/camera/image_raw _camera_frame_id:=v4l_frame. Do not forget the _ before camera_frame_id!)

This will start to look for markers in the image stream, and publish the TF transformation of the detected ones.

The provided launch file lists all of the available parameters. The most important one is the markers configuration. This (YAML) file describes where the markers are on your objects, and lets you publish the position of an object with multiple markers attached. See the sample configuration for a complete example.

• This project is very similar to ROS ALVAR. ALVAR has a broader scope (like support for integration with depth maps), but it is also heavier. Benchmarks have yet to be conducted.
• The standard (but much older) tool is ARToolkit. It also has ROS bindings.