18.07.2014 Views

PHD Thesis - Institute for Computer Graphics and Vision - Graz ...

PHD Thesis - Institute for Computer Graphics and Vision - Graz ...

PHD Thesis - Institute for Computer Graphics and Vision - Graz ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

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

8.3. Localization experiments 145<br />

8.3.2 Path reconstruction<br />

The main application <strong>for</strong> global localization is to compute an initial robot position to initialize<br />

a probabilistic SLAM framework, e.g. [78]. The robot position is then usually maintained by an<br />

extended Kalman filter. The Kalman filter will include the knowledge of previous robot positions<br />

<strong>and</strong> also previous speed <strong>and</strong> heading estimates. Propagating this values probabilistically will<br />

result in a smooth reconstruction of the traversed path.<br />

In absence of a SLAM framework the traversed path has to be reconstructed by global localization<br />

solely. Each position estimate is then computed independently. Figure 8.12 shows a<br />

part of the robots path through the ”Office” environment reconstructed by global localization.<br />

The path consists of 204 independent pose estimated. The pose estimates are marked with<br />

black dots, together <strong>for</strong>ming the traversed path. The red dots mark gross outliers. Table 8.4<br />

summarizes the corresponding numbers. From 204 total pose estimates 21 showed a large deviation<br />

from the original path (from laser localization), thus they are marked as gross outliers.<br />

Such gross outliers would be detected by the Kalman filtering. The average epipolar distance,<br />

computed as a measure of accuracy, is 1.45 pixel.<br />

Figure 8.13 shows the reconstruction of a robots path in the ”Hallway” environment. In fact,<br />

three different sections of a robots path are shown. In total 124 positions have been computed.<br />

The corresponding paths are drawn as black dots. For this scenario no laser ground truth is<br />

available, thus outlier detection did not apply.<br />

#frames (positions) #correct (%) #bad estimates (%) avg. epipolar distance<br />

(std. dev.) [pixel]<br />

Office 204 183 (89.7%) 21 (10.3%) 1.45 (1.0)<br />

Hallway 124 - - 1.49 (0.74)<br />

Table 8.4: Number of correct pose estimates <strong>and</strong> bad estimates <strong>for</strong> the path reconstruction<br />

experiment. The accuracy of the pose estimates are expressed by the epipolar distance.

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

Saved successfully!

Ooh no, something went wrong!