Date of Award

2013

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

First Advisor

Homaifar, Dr. Abdollah

Abstract

The need and rationale for improved solutions to indoor robot navigation is increasingly driven by the influx of domestic and industrial mobile robots into the market. This research has developed and implemented a novel navigation technique for a mobile robot operating in a cluttered and dynamic indoor environment. It divides the indoor navigation problem into three distinct but interrelated parts, namely, localization, mapping and path planning. The localization part has been addressed using dead-reckoning (odometry). A least squares numerical approach has been used to calibrate the odometer parameters to minimize the effect of systematic errors on the performance, and an intermittent resetting technique, which employs RFID tags placed at known locations in the indoor environment in conjunction with door-markers, has been developed and implemented to mitigate the errors remaining after the calibration. A mapping technique that employs a laser measurement sensor as the main exteroceptive sensor has been developed and implemented for building a binary occupancy grid map of the environment. A-r-Star pathfinder, a new path planning algorithm that is capable of high performance both in cluttered and sparse environments, has been developed and implemented. Its properties, challenges, and solutions to those challenges have also been highlighted in this research. An incremental version of the A-r-Star has been developed to handle dynamic environments. Simulation experiments highlighting properties and performance of the individual components have been developed and executed using MATLAB. A prototype world has been built using the WebotsTM robotic prototyping and 3-D simulation software. An integrated version of the system comprising the localization, mapping and path planning techniques has been executed in this prototype workspace to produce validation results.

Download

Share

COinS