deletions | additions       diff --git a/chapter_Introduction_label_chap_introduction__.tex b/chapter_Introduction_label_chap_introduction__.tex  index 5d8ad02..e5bc7a1 100644  --- a/chapter_Introduction_label_chap_introduction__.tex  +++ b/chapter_Introduction_label_chap_introduction__.tex 
...
usually not available in confined spaces and would not provide  sufficiently accurate estimates in cluttered environments.  Processors of MAVs have only limited processing power. It is intended to further reduce the size of MAVs to that of an  insect, hence necessitating lightweight and scalable position  estimation algorithms.  If sufficient computational and physical power is available, a typical  approach to estimate a UAV's position is by using active laser  rangefinders~\cite{grzonka2009towards,bachrach2009autonomous}. 
...
However, this reduced physical payload is not without cost: it must be  traded off against the higher computational payload for the onboard  CPU. Processors of MAVs have only limited processing power.  For example, a standard technique for 3D pose estimation extracts keypoints of the current camera image and a map image and then  determines a homography between both keypoint sets. While this  approach has been used for visual SLAM for