Visualizations in Augmented Reality for Multiple Real-Time Unmanned Aerial Vehicle Video Feeds : Visualizing Occluded and Out-of-View Entities

Sammanfattning: Background. The use of Unmanned Aerial Vehicles (UAVs) is today prevalent in both defense applications as well as in civilian tasks and is predicted to have a major socio-economic impact [22]. The opportunities involved with using Augmented Reality (AR) during UAV operations have engaged previous research intending to improve an operator’s situational awareness which allows operations to be carried out without putting people in dangerous situations [1,11,47]. In the case of a system where multiple UAVs are simultaneously active with the possibility to share information, opportunities arise for visualization of tracked entities either located behind objects in the environment or outside the field of view of a UAV. While previous research has been done in the field of AR on hidden elements, the impact of those techniques in a multi-UAV scenario is unexplored. Objectives. The thesis aimed to answer the question of what the impact is of different visualization techniques in regards to response time, depth perception, and clarity in a multi-UAV scenario involving entities occluded by the environment or otherwise hidden. Methods. This thesis has gone forward with a Systematic Literature Review (SLR) to in a systematic way compile the most important visualization techniques for AR visualizations of hidden elements. The techniques proposed by the SLR were implemented in an application simulating a multi-UAV scenario as well as evaluated in an experiment and a survey to fill in the gaps in previous research. Results. The SLR provided this work with multiple important aspects for visualization of entities, both occluded as well outside the field of view including the importance to utilize occlusion as a depth cue but also concrete techniques for implementation. Through the SLR the Excavation box, Naive ghosting, Silhouette, Ruler, Halo, and Radar techniques were selected for implementation in a multi-UAV scenario. The evaluation showed that for clarity the Excavation box and Silhouette techniques were preferred over the Naive ghosting technique, that the Halo technique is preferred for estimation of position but also aspects such that the occlusion depth cue may be unnecessary at the distances involved in the general multi-UAV scenario evaluated. Conclusions. Based on the findings of this thesis can visualizations in a multi-UAV scenario be created, heightening the situational awareness of UAV operators and therefore allowing tasks to be completed without putting people in dangerous situations [1, 11, 22, 47].