Aiding Navigational Methods with Fiducial Markers : Marker Detection and Localization in 3D Space Using Lidar
Heikkinen, Ville (2025)
2025
Automaatiotekniikan DI-ohjelma - Master's Programme in Automation Engineering
Tekniikan ja luonnontieteiden tiedekunta - Faculty of Engineering and Natural Sciences
Hyväksymispäivämäärä
2025-11-13
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-2025111310605
https://urn.fi/URN:NBN:fi:tuni-2025111310605
Tiivistelmä
Navigation and positioning of different mobile vehicles is important part in many processes. One of them is underground mining. Currently to achieve required accuracy mostly this is done by a total station. Using total station requires an expensive total station and trained operative to use it. It slows down the process and increases the cost.
The proposed solution for this is to use SLAM (Simultaneous localization and Mapping) instead of total station for the navigation. SLAM has its own problems for which fiducial marker detection can solve some of problems. Main problem to solve is to determine the position given by SLAM in larger context (e.g. world coordinates).
Markers with known positions in the world coordinates would be placed inside the mine. If the Markers can be uniquely detected and localized by the SLAM process then their positions would be known in world coordinates and local coordinates used by SLAM. A transformation between the coordinate systems could be calculated. This transformation can then be used to transform any position given by the SLAM to world coordinates, meaning that the navigation of the underground drill rig could be done without a total station.
This thesis focuses on the detection of these fiducial markers with lidar. This includes deciding what kind of marker to use and how it is detected. Because 2D image detection is well studied and it is possible to retrieve intensity image from a lidar this thesis mainly studies if it is possible to use 2D detection techniques with lidar. Also, briefly some ideas about 3D detection are shared.
It is shown that if the used lidar has high enough resolution it is possible to detect an Aruco marker with acceptable accuracy from distance of at least 5 meters. However, the results show that this kind of detection is likely not the best technology to be used. Instead, using 3D detection via model fitting or some other method might yield better results. Also using the lidar intensity values for detecting a reflective surface might also provide benefits. However, it is still shown that it is possible to use 2D detection technologies with only lidar without a camera.
The proposed solution for this is to use SLAM (Simultaneous localization and Mapping) instead of total station for the navigation. SLAM has its own problems for which fiducial marker detection can solve some of problems. Main problem to solve is to determine the position given by SLAM in larger context (e.g. world coordinates).
Markers with known positions in the world coordinates would be placed inside the mine. If the Markers can be uniquely detected and localized by the SLAM process then their positions would be known in world coordinates and local coordinates used by SLAM. A transformation between the coordinate systems could be calculated. This transformation can then be used to transform any position given by the SLAM to world coordinates, meaning that the navigation of the underground drill rig could be done without a total station.
This thesis focuses on the detection of these fiducial markers with lidar. This includes deciding what kind of marker to use and how it is detected. Because 2D image detection is well studied and it is possible to retrieve intensity image from a lidar this thesis mainly studies if it is possible to use 2D detection techniques with lidar. Also, briefly some ideas about 3D detection are shared.
It is shown that if the used lidar has high enough resolution it is possible to detect an Aruco marker with acceptable accuracy from distance of at least 5 meters. However, the results show that this kind of detection is likely not the best technology to be used. Instead, using 3D detection via model fitting or some other method might yield better results. Also using the lidar intensity values for detecting a reflective surface might also provide benefits. However, it is still shown that it is possible to use 2D detection technologies with only lidar without a camera.