Abstract:
The localization of an ambulatory individual, a.k.a. a pedestrian, is a developing domain with the potential to permeate into a variety of applications. Site inspection, which is a prime candidate for time reducing automation, requires localization in a dynamic environment. Building on recent developments in the fields of step detection using Inertial Measurement Units (IMUs) and Structure From Motion (SFM) using a camera rig, this thesis presents research targeted at automating the inspection process. The IMU and camera are attached to the inspector, who transports the unit around key inspection sites, and collects data for processing and gaining a three dimensional model of the structural state. This state can be compared to a theoretical building information model (BIM), permitting the remedy of any errors arising during construction. The focus of this research is the implementation and preliminary testing of localization components. Construction sites, due to their nature, prevent installing additional networks to facilitate pedestrian localization. Since infrastructure based positioning systems are not applicable in such environments; we propose an infrastructure-less solution. It is a twolevel positioning framework, including an Inertial Navigation System (INS) module and an SFM module for the low level localization, and a fusion at the upper level for merging the two techniques. The INS processes an IMU’s output by an Extended Kalman filter (EKF) for error estimation and the integration of many correction techniques such a Zero Updates Velocity (ZUPT), Zero Angular Rate (ZARU), and Heuristic Drift Elimination (HDR). In addition, it applies non-holonomic constraints to the measured motion. As for the high level fusion, inertial and vision data are merged in a total state EKF. The proposed solution is totally infrastructure-less, does not require fingerprinting, and is the first method combining vision and personal dead reckoning with error compensation techniques. Several experiments were conducted t
Description:
Thesis (M.E.)--American University of Beirut, Department of Electrical and Computer Engineering, 2011.
Advisor : Dr. Imad Elhajj, Associate Professor, Electrical and Computer Engineering--Members of Committee : Dr. Ali Chehab, Associate Professor, Electrical and Computer Engineering ; Dr. Daniel Asmar, Assistant Professor, Mechanical Engineering ; Dr. Hiam Khoury, Assistant Professor, Civil and Environmental Engineering.
Includes bibliographical references (leaves 104-108)