deletions | additions       diff --git a/chapter_Introduction_label_chap_introduction__.tex b/chapter_Introduction_label_chap_introduction__.tex  index 14a7150..4919a04 100644  --- a/chapter_Introduction_label_chap_introduction__.tex  +++ b/chapter_Introduction_label_chap_introduction__.tex 
...
flight. In the former case, synthetic images will be generated from  the map image that simulate images taken during flight.  The goal is to estimate $x,y$-coordinates in real-time and on-board of the UAV. It is assumed that the UAV flies at a approximately constant height, such that the estimation of height is not necessary.\section{Contributions}  \label{sec:contributions}  The first contribution of this thesis is a machine learning-based  indoor localization system that is able to run in real-time on board  of an MAV, paving the way to a fully autonomous system. In contrast to  existing \emph{active} approaches, the proposed \emph{passive}  approach only uses a single, monocular, downward-looking camera. Since  computer vision-based localization approaches yield noisy estimates, a  variant of a particle filter was developed that aggregates estimations  over time to produce more accurate predictions. It handles the  estimates of the $k$NN algorithm in an integrative way and resolves  position ambiguities. The method operates on single images and, in  contrast to, for example visual odometry, does not suffer from error  accumulation over time.  The second contribution is a map evaluation technique that predicts  the suitability of a given environment for the proposed algorithm. To  this end, a synthetic data generation tool was developed that creates  random variations of an image. The tool simulates different viewing  angles, motion blur, and lighting settings; the generated synthetic  images are labeled with $x,y$-coordinates based on the 3D position of  the simulated camera model.  The developed software is made publicly available. It encompasses (i)  the localization algorithm as part of the Paparazzi project, which  consists of the texton-based approach in combination with a particle  filter (ii) software for augmenting an image with synthetic views,  (iii) a script for evaluating a map based on histograms and  corresponding $x,y$-positions.