BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Invariant Kalman Filtering - Silvere Bonnabel\, Mines Paris-Tech
DTSTART:20140513T130000Z
DTEND:20140513T140000Z
UID:TALK51770@talks.cam.ac.uk
CONTACT:Tim Hughes
DESCRIPTION:For linear systems\, the well-known Kalman filter has a well-c
 haracterized behaviour that is independent of the underlying system's traj
 ectory in the following sense: the covariance matrix of the estimate\, and
  the Kalman gain\, are identical for all trajectories. On the other hand\,
  for non-linear systems\, due to linearizations around the estimated traje
 ctory\, the extended Kalman filter (EKF) covariance matrix\, gain\, and mo
 re generally behaviour\, do depend on the system's trajectory\, leading to
  possible divergence.\n\nFor non-linear systems possessing symmetries\, th
 e invariant extended Kalman filter (IEKF) is an emerging methodology aimed
  at modifying the conventional EKF so as to account for those symmetries. 
 The resulting filter's behaviour is less (and sometimes not) dependent on 
 the system's trajectory\, leading to improved stability and robustness pro
 perties.\n\nIn this talk\, we will focus on aerospace and mobile robotics 
 applications where the configuration space is a Lie group. The IEKF desig
 n will be detailed\, along with some efficient variants of it\, and some (
 stochastic) stability properties will be discussed. The results will be de
 monstrated experimentally on an attitude estimation problem for a low-cost
  UAV\, and a scan matching SLAM problem for a mobile robot equipped with a
  kinect sensor and odometers.
LOCATION:Cambridge University Engineering Department\, LR6
END:VEVENT
END:VCALENDAR