christianrauch/apriltag_ros

What it solves

This project provides a ROS 2 node that allows robots to detect AprilTags (visual markers) in camera images and determine their exact position and orientation (pose) relative to the camera.

How it works

The node subscribes to rectified images and camera information. It uses the AprilTag library to identify markers in the image and calculates their 3D pose using camera intrinsics and a chosen estimation method (such as PnP). The results are then published as transform data on the /tf topic and as a detailed detection array containing metadata like Hamming distance and decision margins.

Who it’s for

It is designed for developers working with ROS 2 and robotics who need reliable visual fiducial markers for localization, navigation, or object tracking.

Highlights

• Flexible Configuration: Supports multiple tag families (e.g., 36h11, 16h5) and allows for tag-specific sizes and frame names.
• Performance Options: Includes a composable node for efficient intra-process communication to reduce data copying.
• Image Transport Support: Compatible with both raw and compressed image formats.
• Detailed Metadata: Publishes not just the pose, but also the original homography matrix and detection confidence metrics.