------------------------------------------------------------------------------ B O S T O N U N I V E R S I T Y Computer Science Department C O L L O Q U I U M Friday, March 28, 1997 2:00 pm (Coffee served at 1:45 pm, Room MCS 137) Room MCS 135 ------------------------------------------------------------------------------ SPATIAL AGGREGATION: THEORY AND PROGRAMMING LANGUAGE Feng Zhao Intelligent Simulation Group Laboratory for Artificial Intelligence Research Department of Computer and Information Science The Ohio State University Columbus, OH 43210 USA fz@cis.ohio-state.edu http://www.cis.ohio-state.edu/~fz/ We are entering a sensor-rich era: our abilities to sense and act in complex physical environments are increasingly augmented by massive networks of tiny, invisible sensors, actuators, and computers embedded in everything from appliances to materials to building structures. Rapid advances in information processing, micro-fabrication, and MEMS are fueling a new generation of distributed autonomous systems --- fluibots, immobots, and softbots --- in a way resembling how advances in VLSI technology sparked the computer processor revolution in the '80s. However, the computational challenges we are facing are enormous. Much of the information amassed is in the analogue, data-rich form such as continuous streams of images or spatially distributed measurements of physical processes that often have to be consumed in real time. We need to extract macroscopic structures and behaviors from data-massive analogue descriptions so that our reasoning can be articulated in the same way as we explain a picture to a 5-year-old child. We must compile high-level control decisions to actions of local agents so as to cope with the physical environments. Orchestrating distributed computational agents immersed in physical media requires abstraction mechanisms, inference methods, and programming languages different from those for centralized processes. In this talk, I will describe our on-going work in developing theories, languages, and applications of imagistic reasoning and its implementation in the spatial aggregation paradigm. Spatial aggregation builds concise structural and behavioral descriptions from the analogue input so that we (or other programs) can use the compact descriptions to reason about and affect the environments. Spatial aggregation algorithms actively exploit the locality and continuity in the analogue data to gain efficiency and robustness. Spatial aggregation programs are modular and constructed by mix-and-match from a library of commonly used components. We have developed an interpretor SAL/C++ for programs written in this style. If time permits, I will also describe a few engineering applications, including (1) decentralized reasoning and optimization of temperature fields, and (2) a nonlinear, phase-space based maglev control experiment we built in the laboratory. Online papers describing the above work can be found at: http://www.cis.ohio-state.edu/~fz/insight.html. Spatial aggregation is developed in collaboration with Ken Yip. BIOGRAPHY Feng Zhao is currently an assistant professor of Computer and Information Science at the Ohio State University, conducting research in artificial intelligence, scientific computing, and nonlinear dynamics and control. He obtained PhD and MS from M.I.T. Over the last decade, he has developed one of the first linear-time algorithms for the well-know N-body problem, programmed massively parallel computers at Thinking Machines Corporation, worked on MEMS control technology at Xerox Palo Alto Research Center, and served as a consultant to NASA and Hughes Aircraft on advanced information technology. He received the ONR and NSF Young Investigator Awards and is an Alfred P. Sloan Research Fellow in Computer Science. In 1996, he was invited by Japan's Ministry of International Trade and Industry (MITI) to lecture in Japan. Professor Zhao has authored many technical papers and served on program committees for international conferences on artificial intelligence. He currently leads Project Insight, an interdisciplinary Intelligent Simulation Group at Ohio State, with six graduate students and grant support from government and industries. Professor Zhao can be reached at: fz@cis.ohio-state.edu or 614-292-1553. Host: Prof. Stan Sclaroff ------------------------------------------------------------------------------ For colloquium info, including directions, see http://www.cs.bu.edu/colloquium For more information contact Prof. David Yates ------------------------------------------------------------------------------