OpenCV
Library of functions used for computer vision
ArUco Marker
Unique binary marker that a computer can recognize and track through OpenCV
Stereoscopic Vision
Using the distance that an object appears from the left and right side defined as d = x_l - x_r, we were able to calculate several useful parameters that we can use to track an ArUco marker as shown here. One being the distance of the ArUco marker from the camera as well as the horizontal and vertical position of the ArUco marker defined as x and y.
Using only a webcam isn’t sufficient to obtain information about depth, in other words, how far the ArUco marker is relative to the camera, so a second webcam is needed. Using a second webcam, we’re able to determine the distance
Experimental Set Up
Demonstration of the XYZ platform in motion and tracking of the ArUco marker
The ArUco marker move in the following order and direction, and used our ArUco tracking system to follow the position of the ArUco marker while in motion. Additionally, we also wanted to see how the color space of the video impacts how well an ArUco marker can be tracked, so we tested three different color spaces during our experiment. For each of the color spaces, we tested a total of 5 trials allowing us to visualize the impact of ArUco tracking in different light intensity. Especially since the light intensity and color saturation within the operating room can vary which may be potential obstacles for the cameras to track the
Experimental Results
In the table you can see the room mean square error of all 15 experiments. With this data we found that by using inexpensive cameras and set under and RGB color space we can have an error as low as 2mm. Under this color space we also had the lowest amount of filtering. However, going forward we can find alternative methods to further decrease the error of the tracking system and also consider other effects such as the rotating the ArUco marker and movement error.