23.11.2012 Views

Master Thesis - Hochschule Bonn-Rhein-Sieg

Master Thesis - Hochschule Bonn-Rhein-Sieg

Master Thesis - Hochschule Bonn-Rhein-Sieg

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

5. Algorithms <strong>Master</strong> <strong>Thesis</strong> Björn Ostermann page 81 of 126<br />

noise, the size of a region is counted while it is being filling (see chapter 5.2.4.1). If this size is below<br />

a certain threshold, the region is dropped from further processing. Figure 58 shows the objects of the<br />

sample workspace set-up that remain after size filtering.<br />

Figure 58: Remaining objects after region size filter<br />

Objects that only partially enter the camera’s view from the outside cause only small regions. For<br />

example this is the case, if only the hand of a human enters the visual field. If the number of pixels in<br />

those regions is smaller than the threshold, they are eliminated as well. Therefore the application of<br />

this filter is only possible in a workplace, where the camera’s view on the robot is larger than the<br />

robot’s possible reach (see Evaluation of safety in chapter 6.3).<br />

5.2.4.3 Distinguishing the robot from objects<br />

To distinguish the robot from other dynamic objects in the environment, the robot’s base was targeted<br />

manually in the intrusion map (see chapter 3.3).<br />

Every region that is extracted by the region growing algorithm (see chapter 5.2.4.1) is cross referenced<br />

with the base’s centre position. If this position is in the given region, the region is set as the robot’s<br />

region, as shown in Figure 59.<br />

Figure 59: Robot region identified by its base’s centre (blue cross)

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!