Reasoning Over Discontinuous Hypothesis Spaces
Human-Robot Interaction in Complex Applications
Subspace Methods for Vision, Graphics and Signal Processing
Paul Rybski, Aaron Steinfeld, Fernando De la Torre
Carnegie Mellon University
Time and Place
Auditorium (NSH 1305)
To act intelligently in complex and dynamic environments, teams of mobile robots must estimate the position of objects by using information obtained from a wide variety of sources. For any problem of reasonable complexity, teams of mobile robots do not have the sensors necessary to perceive all aspects of a dynamic environment nor correct for unmodeled sources of noise. To address this problem, I propose a method for reasoning over a discontinuous hypothesis space where a strict ordering is imposed on the sources of information. By segmenting the information sources into different classes, a prioritized hierarchy of state estimates are inferred that allows each individual robot to select the most informative state estimate to guide its actions.
There are great opportunities for discovering HRI science within the framework of systems development for complex robotics applications. This talk will provide examples of HRI science during real-world prototype systems deployments and evaluations.
Subspace methods (e.g. Principal Component Analysis, Independent Component Analysis, Linear Discriminant Analysis, ...) have been successfully applied in numerous visual, graphics and signal processing tasks during the last two decades. In this talk, I will provide a unified framework for several novel component-analysis techniques useful for modeling, classifying and clustering huge amounts of high dimensional data. I will discuss how these techniques can be applied to visual tracking, signal modeling (e.g. background estimation) and pattern recognition problems (e.g. face recognition), as well as clustering long term multimodal data (video, audio, body sensors) useful to monitor our daily activity.
Paul E. Rybski (Systems Scientist, Carnegie Mellon University Robotics Institute) received his B.A. in Mathematics/Computer Science with an interdisciplinary emphasis from Lawrence University (Appleton, WI) in 1995. He worked at Honeywell Technology Center in Minneapolis, MN, as a programmer from 1996-1997. He received his M.S. in Computer Science and Engineering from the University of Minnesota in 2001, and then received his PhD from the same department with a minor in Cognitive Science in 2003. While at Minnesota, Paul worked as an independent software contractor for ActivMedia Robotics. Paul joined the RI in the fall of 2003 as a Postdoctoral Fellow working with Professor Manuela Veloso in the multi-robot lab where he has been working on the AIBO and SegwayRMP robot soccer teams. He has also worked on a physical awareness system for meeting understanding as part of the CALO project. He co-chaired the 2005 AAAI Mobile Robot Competition/Exhibition event and will do so again in 2006.
Aaron Steinfeld is a Systems Scientist at the Robotics Institute and member of NavLab. He has extensive experience with human-robot interaction specification, development, and testing for deployed prototype and commercial systems in intelligent transportation, mobile robotics, and military robotics. He has recently brought these skills to the area of multi-agent systems and healthcare robotics.
Fernando De la Torre received his B.Sc. degree in telecommunications, M.Sc. degree in electronic engineering and Ph. D, respectively, in 1994, 1996 and 2001, all from La Salle School of Engineering in Ramon Llull University, Barcelona, Spain. In '98 he did a Master in computer vision at the Computer Vision Center in Barcelona. In '97 and '00 he became assistant and associate professor of the department of Communications and Signal theory in Enginyeria La Salle. During his Ph.D he was a research visitor at Queen Mary and Westfield College, Institute for Advanced Computer Studies- UMIACS at university of Maryland, Xerox Palo Alto Research Center and Brown University. In 2002 Dr. De la Torre was a post-doc at Brown university and Gatsby Neuroscience unit in London. Since 2005 he is research scientist in the Robotics Institute at Carnegie Mellon University.
For appointments, please contact Janice Brochetti.