Tracking in AR

Charlotte Murray
2 min readNov 8, 2020

In AR, tracking is when some objects position and orientation are being continuously measured. This is like face or eye tracking, where it is always being watched. There are three different terms like are used with the alignment and measurement of objects, tracking, calibration, and registration.

Tracking

This is a term to describe the movement sensing and measuring in AR systems. We need to know the relative orientation and position of the AR display and objects compared to the real objects. This must be continuously occurring and updating, so that it operates in real time. This can also be called 3D tracking because it is working with the real world of 3D objects, not 2D like on a computer screen.

Calibration

This is when there are two different devices that must compare their measurements with each other, which will then be aligned to each other. You must determine the parameters for what is called the reference device, to be calibrate with the other device. This needs to be usedon devices that are used for tracking. This usually happens only at certain times. Depending on the device, it may only need to happen once, or everytime it is used.

Registration

This using tracking and calibration to align the cordinate system of the virtual world with the real world. Static registration is when the user or camera is not moving, and this requires calibration. Dynamic registration is when the user or camera is moving, and this using tracking.

Coordinate Systems

These are used often while tracking and calibrating virtual worlds to the reall world. These make it so we can put overlays on top of the real world to enhance it.

Model transformation is the relationship of 3D local object coordinates to the 3D global world coordinates. It is what determines where the virtual objects go in the real world. The view transformation is the relationship of the 3D global world coordinates to the 3D camera coordinates. This is what allows the user to move in the real world during AR scenarios. Projective transformation is the relationship between the 3D camera coordinates and the devices 2D coordinates. This is when your device takes your camera image, and transfer it into a 2D image that you are able to view on your device.

There are different kinds of frames of reference while using AR. There are world stabilized elements which are virtual elements that behaves like a real physical world object. There are also screen stabilized elements that are static to what the user sees, usually in a corner of the users screen. Then there are also body stabilized elements, which will move with the user and aways be the same distance and orientation from the user.

--

--