Investigation of Different Approaches for Visual Odometry for Aerospace Vehicles in Unknown Environments

Date

2023-8

Author

Uzun, Semra Sultan

Metadata

Show full item record

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

Item Usage Stats

22
views

13
downloads

Cite This

Visual sensors have a major advantage over traditional sensing methods for navigation in unknown environments since they provide rich information about the environment. On the other hand, visual measurements cannot be employed directly for accurate localization because they do not provide direct navigation measurement. To extract meaningful information, development of reliable and computationally efficient algorithms are required. This thesis presents two distinct estimation algorithms for angular and translational velocity in unknown environments using the visual measurements. The first algorithm depends on solving the modified Wahba's problem using the optimization techniques and includes the small-angle assumption to estimate the velocities. The second proposed algorithm is based on a Kalman filter with a pseudo measurement model derived from dual quaternion to estimate the velocity vector accurately. Both algorithms are essentially visual odometry methods that provide velocity estimates using the tracked features in the successive images. Their performance is evaluated in different simulation environments and for different aerospace vehicles, which also demonstrates the algorithms are applicable to various platforms. The first test is performed for drone navigation whereas the second one investigates the algorithms' performance for a lunar landing scenario.

Subject Keywords

Visual Odometry, Kalman Filter, Dual Quaternion, Wahba's Problem, Quadrotor, Moon Landing

URI

https://hdl.handle.net/11511/105477

Collections

Graduate School of Natural and Applied Sciences, Thesis