Bender, Daniel: Airborne Navigation by Fusing Inertial and Camera Data. - Bonn, 2018. - Dissertation, Rheinische Friedrich-Wilhelms-Universität Bonn.

Online-Ausgabe in bonndoc: https://nbn-resolving.org/urn:nbn:de:hbz:5n-50464

Online-Ausgabe in bonndoc: https://nbn-resolving.org/urn:nbn:de:hbz:5n-50464

@phdthesis{handle:20.500.11811/7550,

urn: https://nbn-resolving.org/urn:nbn:de:hbz:5n-50464,

author = {{Daniel Bender}},

title = {Airborne Navigation by Fusing Inertial and Camera Data},

school = {Rheinische Friedrich-Wilhelms-Universität Bonn},

year = 2018,

month = apr,

note = {Unmanned aircraft systems (UASs) are often used as measuring system. Therefore, precise knowledge of their position and orientation are required.

This thesis provides research in the conception and realization of a system which combines GPS-assisted inertial navigation systems with the advances in the area of camera-based navigation. It is presented how these complementary approaches can be used in a joint framework. In contrast to widely used concepts utilizing only one of the two approaches, a more robust overall system is realized.

The presented algorithms are based on the mathematical concepts of rigid body motions.

After derivation of the underlying equations, the methods are evaluated in numerical studies and simulations. Based on the results, real-world systems are used to collect data, which is evaluated and discussed.

Two approaches for the system calibration, which describes the offsets between the coordinate systems of the sensors, are proposed. The first approach integrates the parameters of the system calibration in the classical bundle adjustment. The optimization is presented very descriptive in a graph based formulation. Required is a high precision INS and data from a measurement flight. In contrast to classical methods, a flexible flight course can be used and no cost intensive ground control points are required. The second approach enables the calibration of inertial navigation systems with a low positional accuracy. Line observations are used to optimize the rotational part of the offsets. Knowledge of the offsets between the coordinate systems of the sensors allows transforming measurements bidirectional. This is the basis for a fusion concept combining measurements from the inertial navigation system with an approach for the visual navigation. As a result, more robust estimations of the own position and orientation are achieved. Moreover, the map created from the camera images is georeferenced. It is shown how this map can be used to navigate an unmanned aerial system back to its starting position in the case of a disturbed or failed GPS reception. The high precision of the map allows the navigation through previously unexplored area by taking into consideration the maximal drift for the camera-only navigation.

The evaluated concept provides insight into the possibility of the robust navigation of unmanned aerial systems with complimentary sensors. The constantly increasing computing power allows the evaluation of big amounts of data and the development of new concept to fuse the information. Future navigation systems will use the data of all available sensors to achieve the best navigation solution at any time.},

url = {http://hdl.handle.net/20.500.11811/7550}

}

urn: https://nbn-resolving.org/urn:nbn:de:hbz:5n-50464,

author = {{Daniel Bender}},

title = {Airborne Navigation by Fusing Inertial and Camera Data},

school = {Rheinische Friedrich-Wilhelms-Universität Bonn},

year = 2018,

month = apr,

note = {Unmanned aircraft systems (UASs) are often used as measuring system. Therefore, precise knowledge of their position and orientation are required.

This thesis provides research in the conception and realization of a system which combines GPS-assisted inertial navigation systems with the advances in the area of camera-based navigation. It is presented how these complementary approaches can be used in a joint framework. In contrast to widely used concepts utilizing only one of the two approaches, a more robust overall system is realized.

The presented algorithms are based on the mathematical concepts of rigid body motions.

After derivation of the underlying equations, the methods are evaluated in numerical studies and simulations. Based on the results, real-world systems are used to collect data, which is evaluated and discussed.

Two approaches for the system calibration, which describes the offsets between the coordinate systems of the sensors, are proposed. The first approach integrates the parameters of the system calibration in the classical bundle adjustment. The optimization is presented very descriptive in a graph based formulation. Required is a high precision INS and data from a measurement flight. In contrast to classical methods, a flexible flight course can be used and no cost intensive ground control points are required. The second approach enables the calibration of inertial navigation systems with a low positional accuracy. Line observations are used to optimize the rotational part of the offsets. Knowledge of the offsets between the coordinate systems of the sensors allows transforming measurements bidirectional. This is the basis for a fusion concept combining measurements from the inertial navigation system with an approach for the visual navigation. As a result, more robust estimations of the own position and orientation are achieved. Moreover, the map created from the camera images is georeferenced. It is shown how this map can be used to navigate an unmanned aerial system back to its starting position in the case of a disturbed or failed GPS reception. The high precision of the map allows the navigation through previously unexplored area by taking into consideration the maximal drift for the camera-only navigation.

The evaluated concept provides insight into the possibility of the robust navigation of unmanned aerial systems with complimentary sensors. The constantly increasing computing power allows the evaluation of big amounts of data and the development of new concept to fuse the information. Future navigation systems will use the data of all available sensors to achieve the best navigation solution at any time.},

url = {http://hdl.handle.net/20.500.11811/7550}

}