This paper presents a first attempt to use colored pixel segmentation for mobile robot navigation and localization. Because of the wide variety of colors that are present in a general semi-structured environment, the non-supervised approach is selected. In order to obtain a robust classification valid for different lighting conditions, we also try different color/brightness parameters combinations to select a better variable set than a single color-space can offer. Two learning paradigms are tried, evaluated and compared: a Self Organizing Map (SOM) and Forgy's clustering algorithm. A new performance evaluation measure is also presented, in order to select the best method/feature subset combination that will give better classification results.
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.