| dc.description.abstract | This thesis documents the development of a system for estimating the position of avalanche victims using unmanned aerial vehicles (UAV).
The system can only be used if the buried individual is wearing a transmitting avalanche beacon compatible with the standard of design for such equipment. The estimation process is based on the Sum of Gaussians estimation technique for estimating the state of non-linear systems with multi modal covariance.
This technique utilizes multiple parallel extended Kalman Filters working on a set of Gaussian distributions based on the current measurements made by the system. The final estimate is computed by assigning and altering weights for each Gaussian. The development of a prototype UAV based on off-the-shelf hardware has also been documented. Two programs have been developed in order to simulate measurements on board the UAV and transmit these to a computer running the estimation program. The project is based on open-source software.
The thesis has been completed in cooperation with Options AS and co-supervisor Espen Møller which has provided the necessary components for a prototype UAV. Insight about system design and the idea and specification for the project has been provided by Options. Several discussions with Espen about the general software system design has also eased the implementation.
The master thesis is a continuation of specialization course TTK4555 completed in the fall of 2017. During this course the workings of the avalanche transceiver was examined and different estimation techniques researched. A crude implementation of the Sum of Gaussian estimation was done as a part of this course.
Rune Storvold is the main supervisor of the thesis. General information about UAVs and use of these was provided by Rune during the specialization course.
During the writing of this thesis and the implementation of the system, no specific background information, starting point or equipment were provided. Help with specific problems regarding hardware or the libraries used has not been provided, as the company has no specific competence on this. The thesis is based on an idea by Options and the prototype realized with the help of their funding.
A working UAV has been built and a program for interfacing with the flight controller on-board this developed. A program running the estimation of the buried avalanche transceiver has also been developed. The UAV have been flight tested and the estimation software has been tested both with simulated measurement positions and with position data from the UAV.
It was not possible to obtain an actual sensor system with capability to share its measurement data during this thesis. Building a prototype sensor system was researched and considered, but found to be too time-consuming and likely to fail.
The initial idea was to utilize multiple UAVs to generate measurements from different positions. The estimation software is developed with this in mind and supports multiple data sources. However, only one prototype UAV was built as handling and piloting multiple UAVs at once by one person may be in conflict with the current regulations for UAVs. This would also prolong the time required to build the UAVs, increase the cost of the project and make transportation for testing difficult. | |
