deletions | additions       diff --git a/chapter_Conclusion_label_chap_conclusion__.tex b/chapter_Conclusion_label_chap_conclusion__.tex  index 8d183b8..4a5191d 100644  --- a/chapter_Conclusion_label_chap_conclusion__.tex  +++ b/chapter_Conclusion_label_chap_conclusion__.tex 
...
\chapter{Conclusion}  \label{chap:conclusion}  This thesis presented a novel approach for fast and accurate lightweight  indoor localization of MAVs. We pursued an onboard design without the need of  an additional ground station to foster flexibility and autonomy. The  conducted on-ground and in-flight experiments underline the real-world  applicability of the system. Promising results were obtained for  waypoint navigation, accurate landing, position estimates  and stable hovering accurate landing  in the indoor environment. Theused  approach is based on three pillars.\\ pillars that we identified for indoor  localization for MAVs.  The first pillar shifts computational effort from the flight phase to an offline preprocessing step. This allows for using provides  the advantages of  sophisticated algorithms, without affecting performance during flight. The second pillar states that during flightthe MAV lightweight  algorithms should run with low-performing processors. These  algorithms should be able to used that can  trade off speed with accuracy. This allows to their  usethem  on a wide range of models, from pico-drones to MAVs with a wing-span of over one meter. Examples of these adaptable algorithms are the particle filter texton-based approach  and the texton-based approach. particle filter.  The third pillar is the a  known---and possibly---modifiable environment. This knowledge and flexibilty flexibility  allows to predict for predicting and improving  the quality of theused  approach. In %In  contrast to SLAM \textsc{slam}  frameworks, in which %which  the task is to simultaneous mapping and localization, the proposed %presented  approach is intended for various repetitive indoor activities. %activities.