Working with AR Marker detection for object pose estimation

In this section, we will see how to use fiducial markers in order to enable a robot to easily interact with its environment. To interact with arbitrary objects, a robot should be able to recognize and localize them by relying on its vision sensors. Estimating the pose of an object represents an important feature of all robotic and computer-vision applications. However, efficient algorithms to perform object recognition and pose estimation working in real-world environments are difficult to implement, and in many cases one camera is not enough to retrieve the three-dimensional pose of an object.

More precisely, with the use of only one fixed camera, it is not possible to get spatial information about the depth of a framed scene. For this reason, object pose estimation is often simplified by exploiting AR markers. An AR marker is typically represented by a synthetic square image composed by a wide black border and an inner binary matrix...