Submitted version of the thesis - Airlab, the Artificial Intelligence ...
Submitted version of the thesis - Airlab, the Artificial Intelligence ...
Submitted version of the thesis - Airlab, the Artificial Intelligence ...
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Chapter 5<br />
Vision<br />
The vision system is focused on <strong>the</strong> implementation <strong>of</strong> <strong>the</strong> camera for object<br />
detection and calculating <strong>the</strong> position <strong>of</strong> this object according to <strong>the</strong> robot.<br />
To do so, <strong>the</strong> relation between different coordinate systems, which are image<br />
planecoordinate system, robot coordinatesystem, andreal world coordinate<br />
system, should be defined. In order to go to <strong>the</strong> target, it is necessary to<br />
know where <strong>the</strong> object is placed. Using <strong>the</strong> blob search algorithm, we acquire<br />
<strong>the</strong> position <strong>of</strong> <strong>the</strong> object in <strong>the</strong> pixel coordinates. We need a mapping<br />
between those three coordinates to command <strong>the</strong> movement. There are several<br />
approaches to determine <strong>the</strong> object position according to <strong>the</strong> robot, as<br />
mentioned in Chapter 2. Among those, we will use <strong>the</strong> one for <strong>the</strong> calibrated<br />
cameras.<br />
The rest <strong>of</strong> Chapter describes camera calibration, <strong>the</strong> script that calculates<br />
<strong>the</strong> transformation, color definition, and <strong>the</strong> script to define <strong>the</strong> color.<br />
5.1 Camera Calibration<br />
The initial step for <strong>the</strong> algorithm is to find <strong>the</strong> camera calibration. The<br />
Camera Calibration Toolbox [2] is used to calculate <strong>the</strong> camera calibration<br />
matrix that is necessary to find <strong>the</strong> projection between <strong>the</strong> world and image<br />
points represented by homogeneous vectors. Normally, <strong>the</strong> object is placed<br />
on <strong>the</strong> real world coordinate systems, and its position can be measured from<br />
<strong>the</strong> origin. In order to find <strong>the</strong> distance <strong>of</strong> <strong>the</strong> camera plane to <strong>the</strong> object,<br />
we need a transformation that maps <strong>the</strong> 3D points (for <strong>the</strong> object that is<br />
assumed to place in ground, <strong>the</strong> point is a 2D point in world coordinates,<br />
since height is 0 all <strong>the</strong> times) <strong>of</strong> <strong>the</strong> objects in <strong>the</strong> 3D world coordinates to