Apriltag Tracker

The following is code that will retrieve planar coordinates of a camera using an apriltag.

Dependencies

This code requires the apriltag module, numpy and OpenCV 3.4.1

Figuring Out The Experimental Factor

As described in the documentation for the init function for the ApriltagTracker class, you will need to calculate an experimental constant to scale coordinates calculated by the object into real-world units. This will vary from camera-to-camera, but can be easily found by following these steps:

1. Place the camera directly in front of the apriltag.
2. Set "exp_factor" in the code to 1.
3. Run the code, and record both the distance the camera is from the apriltag as well as the calculated y coordinate (the x coordinate should be approximately zero)
4. Repeat this for a number of different distances, but remain directly in front of the tag.
5. Calculate the ratio between each distance and its correspond y coordinate.
6. Set "exp_factor" to the average of these ratios.

How To Use

To use the module, you need to do two things within your code. First you need to start a thread that runs the "takePics" function. This will populate the image queue.

The other thing you need to do, is instantiate a tracker object with the appropriate values. This object is what will actually be processing the images to extact the pose.

All you need to do is call the analyzeFrame function everytime you want to get the pose of from the most recently taken picture.

If you have a series of images you want to process that are not taken by a camera in real-time, this is possible as well. Just make sure you never call the "takePics" function, and instead you pass each image into the analyzeFrame function.