| In this thesis modern vision based localisation methods are discussed and contrasted with existing satellite based approaches. Shortcomings are noted and potential solutions are highlighted. A novel method of using a gimballed camera to perform visual Simultaneous Localisation and Mapping (SLAM) is proposed, along with a control algorithm to point the camera toward feature dense regions. This method is then modularly coupled with existing visual SLAM techniques allowing seamless integration across different platforms. Ground tests are performed to verify operation of the gimbal controller and rotation inverser. Results from experimental flight tests are incorporated as a final means of obtaining information to verify gimbal operation. |
