MSc Dissertation: Shuaib Omar


Omar, Shuaib. Target Tracking using Multisensor Data Fusion for an Unmanned Aerial Vehicle Sense and Avoid System. MSc Dissertation. Department of Electrical Engineering, University of Cape Town, 2012.


The last decade has seen a proliferation of unmanned aerial vehicles (UAV) put to use in a myriad of applications. However, the true potential of UAVs lies in its timely adoption to civil airspace. For this to occur, UAVs need to exhibit increased levels of autonomy. The principal impediment in achieving this autonomy is the UAV‘s capability to sense and avoid. The aim of this dissertation is to develop an airborne threat tracking system using multisensor data fusion for a UAV sense and avoid system, termed the threat tracking unit (TTU).

This thesis entails an extensive study of relevant literature in the fields of target tracking, estimation and multisensor data fusion. Following this, a simulation environment was developed to test and qualify various iterations of the TTU.

Initial TTU designs entailed the design of single sensor tracking filters in order set a baseline with which the performance of multisensor tracking filters can be compared against. Subsequent to this, multisensor fusion architectures were investigated drawing on the shortcomings of their single sensor counter-parts.

A centralised fusion architecture (FA1) demonstrated superior performance in both position and velocity tracking. However, this architecture possesses a single point of failure in flight-critical situations. Comparatively, a distributed fusion architecture (FA2) displayed adequate performance. This architecture, nevertheless, is characterised by increased robustness and lower computational load.

Finally, in order to further reduce computational complexity, a full dynamic feedback fusion architecture (FA3) was developed. Although FA3 displayed reduced tracking accuracy, feasible results were still obtained.