This paper presents a self-organizing intelligent fuzzy system for position estimation of a mobile robot based on vision data. The proposed method does not use either landmarks or artificial symbols. An omnidirectional local vision system was developed for the experiments in order to offer the images that reflect the environment. Three techniques for image features extraction are discussed, compared and used in the experiments: Curvilinear Component Analysis, Principal Component Analysis and Output Related Features. The results show that ORF and CCA techniques offered the best features for position estimation of the robot. ln a subspace spanned by the feature vectors, a self-organizing B-spline fuzzy controller is applied by covering the features obtained from the dimension reduction phase with linguistic terms. All the experiments were realized in a natural office environment, even when the light intensity varied, the results emphasizing the robustness of the self-organizing intelligent fuzzy system proposed by this paper.
Self-organizing fuzzy system, B-spline fuzzy controller, features extraction from image vectors.
All articles published by Studies in Informatics and Control (SIC) are published in accordance with the terms and conditions of the Creative Commons Attribution License (CC BY-NC-SA 4.0).