3D pose uncertainty estimation for visual aided navigation

Date

2024-12-5

Author

Temür, Ramazan

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

138
views

65
downloads

Cite This

In Simultaneous Localization and Mapping (SLAM) technologies, it is crucial to navigate unknown environments by utilizing sensor measurements. Navigation systems consist of different types of sensors, such as IMUs, GNSS receivers, LIDAR, and cameras. Fusing this range of sensors ensures precision and safety up to a level; however, accuracy still needs to be improved due to the potential errors of the sensors and the algorithms used in navigation calculations. This study investigates uncertainty quantification in 3D pose estimation algorithms for visual-aided navigation systems. The analytical uncertainty derivations for pose estimation will be obtained to be aware of the correctness of the results before fusing the outputs of homography and the 8-point algorithm. The uncertainties in feature location will be analyzed based on different feature extraction methods to employ the uncertainty of image points to these algorithms. The other focus of this study is to reveal how pose estimation outcomes are influenced by modeling the feature errors as isotropic or anisotropic. Consequently, incorporating error characteristics into pose estimation algorithm frameworks will enhance the results and provide a performance metric to be used in real-world navigation algorithms.

Subject Keywords

Computer vision, Uncertainty, Navigation, 3D pose estimation

URI

https://hdl.handle.net/11511/112931

Collections

Graduate School of Natural and Applied Sciences, Thesis