Aristotle University Of Thessaloniki (AUTH)
Ταυτόχρονη χαρτογράφηση και εντοπισμός με τεχνικές ευφυών συστημάτων
Abstract
dc:descriptionThe current doctoral thesis deals with the problem of the Simultaneous Localization And Mapping known as SLAM. During the research two new and novel methods were proposed, the L-SLAM and the G-SLAM. Both methods use the probabilistic approach of the problem. L-SLAM is based on the observation that almost all kinematic models can be separated into two parts, alinear and a non linear. Exactly like the FastSLAM methods in L-SLAM the first is estimated with particle filters and the second with Kalman filters. The novelty of our method is that only the non linear part of the kinematic model is estimated using particle filters, while the remaining linear part is estimated with Kalman filters. In contrast with FastSLAM which deals the whole kinematic model with particle filters, in L-SLAM using only a part of it we reduce its dimensionality and therefore its complexity from O^3 to O^1 for 2D problems. Also in L-SLAM linear Kalman filters are used, while in FastSLAM it is used non linear Extended Kalman filters. Also it is described the extension of the L-SLAM from the 2D to 3D environments. In this situation the dimensionality reduction is bigger and the subsequent complexity is reduced from O^6 to O^3.The second method we have developed is G-SLAM. In contrast to the previous methods, G-SLAM produces maps of a much better resolution, a very important capability for indoor mobile robots. Many high resolution methods were proposed in the robotics literature. Most of them are based on the Grid Occupancy SLAM. G-SLAM is based on the observation that all of these methods are calculating over and over meaningless point on the map. In G-SLAM the map consists of a scattered points set which using G-SLAM technique tends to gather around obstacles, producing a high resolution map with much less amount of map points.Finally it is presented experimental results from all methods using simulated dataset and datasets from real case scenarios.
Degree
thesis:*- Grantor dc:publisher
- Aristotle University Of Thessaloniki (AUTH)
- Year dc:date
- 2014
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Zikos, Nikolaos
- Ζήκος, Νικόλαος
Subjects
dc:subject × 13- Χαρτογράφηση
- Εντοπισμός
- Particle φίλτρα
- Μείωση διαστάσεων
- Mapping
- Localization
- SLAM
- Particle filters
- Dimensionality reduction
- Επιστήμες Μηχανικού και Τεχνολογία
- Επιστήμη Ηλεκτρολόγου Μηχανικού, Ηλεκτρονικού Μηχανικού, Μηχανικού Η/Υ
- Engineering and Technology
- Electrical Engineering, Electronic Engineering, Information Engineering
Rights
dc:rights- Statement dc:rights
-
- BY
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/39927