A Monocular Vision-based System Using Markers for a Real-Time 6D Pose Estimation of a Trailer
Author
Term
4. semester
Education
Publication year
2022
Submitted on
2022-06-01
Pages
56
Abstract
This project investigates the possibilities in object detection and pose estimation with a single monocular camera without the use of neural networks. The object in need of localization is a trailer. This report dives into current marker design along with current methods for detecting these designs. Furthermore, this report describes different sensor systems, along with the advantages and disadvantages of them to compare. Rather than detecting the object itself, a marker has been created to attach onto the object. The marker consists of two spheres of the color blue and orange. Detecting the marker, robotic perception method has been implemented, such as Hough circle detection and color segmentation. The pose estimation system works within a distance of maximum 560mm, when both of the spheres are in the camera frame. Last, the unexpected and expected test results have been discussed, along with thoughts upon future work.
Keywords
Documents