Your browser does not support JavaScript!

Home    Προβολικές-Μεταθετικές Αναλλοίωτες ως Οπτικά Ορόσημα του Χώρου για Τοπολογική Πλοήγηση  

Results - Details

Add to Basket
[Add to Basket]
Identifier uch.csd.msc//1998tsonis
Title Προβολικές-Μεταθετικές Αναλλοίωτες ως Οπτικά Ορόσημα του Χώρου για Τοπολογική Πλοήγηση
Alternative Title Autonomous Landmark Based Navigation using Projective and Permutation Invariants
Creator Tsonis, Vassilios S
Abstract Autonomous robot navigation is one of the most challenging problems in robotics. A promising approach to this problem is based on the use of landmarks, i.e. distinctive visual events which can be easily recognized by the robot and used for its self-localization in space. However, the complexity of landmark identification and recognition usually leads to assumptions concerning the existence of previous knowledge regarding the robot workspace as well as the objects used as landmarks. Therefore, such an approach cannot be easily applied in dynamically changing environments. In this thesis we propose a method for indoor navigation using visual landmarks that is not based on {\em a priori} knowledge of the environment. Landmark-based navigation usually relies on the identification and subsequent recognition of a number of environment objects, that are deemed adequate in describing the workspace structure. This process is inherently difficult in practice, due to the many different poses of an object that may be encountered in navigational trials. In our method we overcome this problem by employing projective invariants computed on quintuples of points, which are considered as workspace landmarks. Such quantities remain invariant under different camera positions and provide for effective description of the workspace structure. In order to identify potential corresponding quintuples in image frames, we introduce a simple test based on the covariance matrix estimate of each quintuple. With this test, we effectively by-pass the calculation of point correspondences. Since the above test indicates correspondence between quintuples, and not between their individual points, we subsequently employ a permutation insensitive projective invariant for quintuple recognition. Prerequisite to the navigation procedure is a learning phase. In this phase the robot extracts landmarks from the workspace. These landmarks are stored in the robot's database, along with information regarding their relative position in space. As a result, a topological map of the workspace is constructed, based on the extracted landmarks. This map is consulted later, during a navigation session, in order to decide the next motion(s). During navigation, the robot compares the landmarks that it encounters with the ones stored in memory. Successful recognition of a landmark allows the robot to approximately localize itself and select its next movement. Our approach has been evaluated using image sequences from indoor corridors. The results obtained are quite promising and serve as an indication of the method's effectiveness.
Issue date 1998-11-01
Date available 1998-11-24
Collection   School/Department--School of Sciences and Engineering--Department of Computer Science--Post-graduate theses
  Type of Work--Post-graduate theses
Views 479

Digital Documents
No preview available

Download document
View document
Views : 5

No preview available

View document