ABSTRACT: In recent years, both academic and industrial research have focused a huge effort on studying the distributed system paradigm and a wide variety of related applications. Many of these applications are based on the design of (relatively) cheap devices, characterized by computational and communication capabilities, memory, and instrumented with sensors and/or actuators to interact with the peers and the surroundings: a network of such devices is called a multi-agent systems. It is remarkable how, in many situations, the multi-agent system shows emergent behaviors and performances that are higher than those of the monolithic architecture counterpart. At the same time, though, this new perspective poses multi-disciplinary issues that span over the communication, computer science, system theory fields of research (to cite a few). In this talk, some of these problems are presented, such as those regarding the observability and the topology understanding in sensor networks, and the finite resource problem in an heterogeneous agent multi task framework. Although of general discussion and application, we will refer to the specific case study of a camera network for environment monitoring and surveillance, which, beyond the importance per se, appears also as an appealing paradigmatic example. In this context, we will first formalize three problems, namely the "graph building problem", the "task assignment problem", and the "cooperation problem" and then propose some algorithmic solutions, which have been validated with numerical simulations using real world scenario indications and parameters.
BIO: Angelo Cenedese (Laurea 1999, Ph.D 2004) is currently Assistant Professor at the Department of Engineering and Management of the University of Padova (Italy). He has held several visiting positions with international institutions: the JAERI Institute, Naka, Japan (2000), the UKAEA-JET laboratory, Oxford, UK (2001-2004), General Atomics, SanDiego, CA (2004), UCLA, Los Angeles, CA (2010). He has been and he is currently involved in EU Projects on control and diagnostics of nuclear fusion devices, on methodologies for Adaptive Optics systems, on estimation and control in distributed networked systems. His interests are in the fields of modeling, control theory and its applications, active vision, sensor and actor networks, with particular
attention to environmental monitoring and control, and surveillance networks. He is co-author of more than 40 papers, and three international patent applications on the field of surveillance sensor networks are under
HOST: Francesco Bullo