Andrei Stancovici, Public Dissertation of PhD Thesis
Thesis Title: "Relative Localization Methodology in Collaborative Robotic Environments"
Author: MSc. Eng. Andrei STANCOVICI
- Chair: Professor Dr. Eng. Radu-Emil PRECUP (Politehnica University of Timisoara)
- PhD Supervisor: Professor Emeritus Dr. Eng. Vladimir I. CREŢU (Politehnica University of Timisoara)
- Scientific Referees:
- Professor Dr. Eng. Sergiu NEDEVSCHI (Technical University of Cluj-Napoca)
- Professor Dr. Eng. Nicolae ŢĂPUŞ (University Politehnica of Bucharest)
- Professor Dr. habil. Eng. Mihai V. MICEA (Politehnica University of Timisoara)
This paper addresses the problem of localization in the context of exploring unknown environments by autonomous robots.
In order to solve this problem and due to the lack of a standardization of the localization research in the interior environment, the author proposes a relative localization methodology for autonomous robotic systems with a cooperative role.
The methodology is applicable for a wide range of systems. In this paper, the CORE-TX system (COllaborative Robotic Environment – the Timisoara eXperiment) was taken as an example. There are three levels of localization offered by the methodology: PREDICTION, COOPERATION and CENTRALIZATION.
For the COOPERATION level, a mathematical model based on the ILS (Iterative Least Squares) method is proposed, which can solve the localization problem without the need of a central computing equipment. The CENTRALIZATION level offers a superior confidence level based on high-power central equipment and an algorithm for distributed processing systems. The BPF (Backtracking Particle Filter) proposed algorithm, with an original localization technique, is a "particle filter" algorithm in
a form of "backtracking" and which is based on the Las Vegas probability algorithms.
A hardware device called IRULT (Inter-Robot Ultrasonic Localization Turret) has been developed with an original concept that is positioned on a turret and is equipped with two ultrasonic transducers to get the robotic orientation and position in the navigation task.
Because the experiments in practice cover a too isolated case to validate the proposed methodology, several sets of experiments are being run through simulation to cover all isolated cases that may occur in practice.
By analysing the experimental results and introducing the "base" concept with the role of resetting the propagation errors in localization, it is shown that the proposed relative localization methodology is applicable in the context of exploring unknown environments by autonomous robots.