MPhil Thesis Defence


"A Location-guided Mobile Robot Control Platform"

By

Mr. Ran Liu


Abstract

Driven by the increasing demand for mobility and automation in both
research and application domains, mobile robots have started to draw wide
attention from researchers in academia as well as industry. Among the
innumerable research topics in robotics, two fundamental issues lie in the
technologies of location sensing and path planning. This thesis studies
the two issues in real world environments which are inherently uncertain
and dynamic.

We address the location sensing issue with an active Radio Frequency
Identification (RFID) based indoor location sensing technology. The
technique uses adaptive temporal radio maps and reference points to
enhance computation efficiency and offset environmental variations. In
particular, we investigate the RFID tag signal characteristics in an
indoor environment and analyze the feasibility of utilizing Active RFID
for location estimation. Then, we propose a clustering technique and
probabilistic methodology to achieve higher estimation accuracy and
improved systematic properties including maintainability, scalability and
cost.

To address the issue of planning with imperfect information in dynamic
environments, we propose a heuristic-based, anytime planning and
replanning algorithm. Our algorithm combines the benefits of anytime and
incremental planning to provide efficient solutions to complex, dynamic
search problems. Besides, obstruction detection is achieved by observing
the patternized signal strength variation of a certain type of obstruction
while the obstruction avoidance is taken care of by the anytime replanning
within the available deliberation time.

A case study of the research results is conducted with a prototype system.
The system is built utilizing composition methodology. The location
sensing and path planning subsystems are implemented as components such
that they can operate and interoperate with the component-oriented
software infrastructure through plug-and-play integration. The case study
aims to provide qualitative analysis of the scientific methodologies and
real world values behind the prototype as well as demonstrating its
functionality and applicability in learning activities.


Date:				Monday, 15 August 2005

Time:				4:00p.m.-6:00p.m.

Venue:				Room 1504
				Lifts 25-26

Committee Members:		Prof. Lionel Ni (Supervisor)
                        	Dr. Yunhao Liu (Chairperson)
                        	Dr. Jogesh Muppala


**** ALL are Welcome ****