This thesis is concerned with the Distributed Information Gathering (DIG) problem in which a Sensor Network is tasked with building a common representation of environment. The problem is motivated by the advantages offered by distributed autonomous sensing systems and the challenges they present. The focus of this study is on Macro Sensor Networks, characterized by platform mobility, heterogeneous teams, and long mission duration. The system under consideration may consist of an arbitrary number of mobile autonomous robots, stationary sensor platforms, and human operators, all linked in a network. This work describes a comprehensive framework called Active Sensor Network (ASN) which addresses the tasks of information fusion, decistion making, system configuration, and user interaction. The main design objectives are scalability with the number of robotic platforms, maximum flexibility in implementation and deployment, and robustness to component and communication failure. The framework is described from three complementary points of view: architecture, algorithms, and implementation. The main contribution of this thesis is the development of the ASN architecture. Its design follows three guiding principles: decentralization, modularity, and locality of interactions. These principles are applied to all aspects of the architecture and the framework in general. To achieve flexibility, the design approach emphasizes interactions between components rather than the definition of the components themselves. The architecture specifies a small set of interfaces sufficient to implement a wide range of information gathering systems. In the area of algorithms, this thesis builds on the earlier work on Decentralized Data Fusion (DDF) and its extension to information-theoretic decistion making. It presents the Bayesian Decentralized Data Fusion (BDDF) algorithm formulated for environment features represented by a general probability density function. Several specific representations are also considered: Gaussian, discrete, and the Certainty Grid map. Well known algorithms for these representations are shown to implement various aspects of the Bayesian framework. As part of the ASN implementation, a practical indoor sensor network has been developed and tested. Two series of experiments were conducted, utilizing two types of environment representation: 1) point features with Gaussian position uncertainty and 2) Certainty Grid maps. The network was operational for several days at a time, with individual platforms coming on and off-line. On several occasions, the network consisted of 39 software components. The lessons learned during the system's development may be applicable to other heterogeneous distributed systems with data-intensive algorithms.

Thesis (M.A. Sc.)--University of Toronto, 2004. / Adviser: G.M.T. D'Eleuterio.

Thesis (M.C.S.) - Carleton University, 2005. / Includes bibliographical references (p. 153-161). Also available in electronic format on the Internet.

Thesis (doctoral)--Helsinki University of Technology, 2003. / Includes bibliographical references (p. 136-143). Also available on the World Wide Web.

Thesis (Ph.D)--Computing, Georgia Institute of Technology, 2009. / Committee Chair: Dellaert, Frank; Committee Member: Bobick, Aaron; Committee Member: Christensen, Henrik; Committee Member: Leonard, John; Committee Member: Rehg, James. Part of the SMARTech Electronic Thesis and Dissertation Collection.

Thesis (Ph.D)--Computing, Georgia Institute of Technology, 2010. / Committee Chair: Arkin, Ronald C.; Committee Member: Christensen, Henrik I.; Committee Member: Fisk, Arthur D.; Committee Member: Ram, Ashwin; Committee Member: Thomaz, Andrea. Part of the SMARTech Electronic Thesis and Dissertation Collection.

Thesis (Ph. D.)--University of Cambridge, 1991. / "November 1991." Includes bibliographical references.

Thesis (Ph. D.)--Electrical and Computer Engineering, Georgia Institute of Technology, 2006. / Ashraf Saad, Committee Member ; Spyros Reveliotis, Committee Member ; Anthony Yezzi, Committee Member ; Erik Verriest, Committee Member ; Yorai Wardi, Committee Co-Chair ; Magnus Egerstedt, Committee Chair.

Thesis (Ph. D.)--Oregon State University, 2010. / Printout. Includes bibliographical references (leaves 134-142). Also available on the World Wide Web.

In this dissertation, we present and evaluate algorithms for statistical gas distribution modelling in mobile robot applications. We derive a representation of the gas distribution in natural environments using gas measurements collected with mobile robots. The algorithms fuse different sensors readings (gas, wind and location) to create 2D or 3D maps. Throughout this thesis, the Kernel DM+V algorithm plays a central role in modelling the gas distribution. The key idea is the spatial extrapolation of the gas measurement using a Gaussian kernel. The algorithm produces four maps: the weight map shows the density of the measurements; the confidence map shows areas in which the model is considered being trustful; the mean map represents the modelled gas distribution; the variance map represents the spatial structure of the variance of the mean estimate. The Kernel DM+V/W algorithm incorporates wind measurements in the computation of the models by modifying the shape of the Gaussian kernel according to the local wind direction and magnitude. The Kernel 3D-DM+V/W algorithm extends the previous algorithm to the third dimension using a tri-variate Gaussian kernel. Ground-truth evaluation is a critical issue for gas distribution modelling with mobile platforms. We propose two methods to evaluate gas distribution models. Firstly, we create a ground-truth gas distribution using a simulation environment, and we compare the models with this ground-truth gas distribution. Secondly, considering that a good model should explain the measurements and accurately predicts new ones, we evaluate the models according to their ability in inferring unseen gas concentrations. We evaluate the algorithms carrying out experiments in different environments. We start with a simulated environment and we end in urban applications, in which we integrated gas sensors on robots designed for urban hygiene. We found that typically the models that comprise wind information outperform the models that do not include the wind data.