BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Quasi-linear Sensor Management - Marco Huber (University of Karlsr
 uhe)
DTSTART:20090306T150000Z
DTEND:20090306T160000Z
UID:TALK17380@talks.cam.ac.uk
CONTACT:Zoubin Ghahramani
DESCRIPTION:In multimodal sensor systems\, such as sensor networks\, the m
 ain objective of sensor management is to utilize sensors and their sensing
  modalities in a most informative way. This requires planning decisions ov
 er multiple time steps. For computationally constrained sensor systems\, t
 he “quasi-linear sensor management” approach allows efficient and mean
 ingful decision making for nonlinear non-Gaussian systems by utilizing ope
 n-loop feedback control. In this talk\, three key techniques of the quasi-
 linear sensor management approach are presented: (1) predictive statistica
 l linearization for converting nonlinear non-Gaussian sensor management pr
 oblems into linear Gaussian ones\, (2) optimal branch-and-bound pruning fo
 r quickly solving linear Gaussian sensor management problems\, and (3) Gau
 ssian mixture reduction for bounding the growth of the number of mixture c
 omponents when employing Gaussian mixture models for Bayesian filtering. T
 he effectiveness of these techniques is demonstrated by means of simulatio
 ns including mobile sensor control and sensor scheduling in sensor network
 s.\n
LOCATION:Engineering Department\, CBL Room 438
END:VEVENT
END:VCALENDAR