Group for Research in Decision Analysis

State of the art in Gaussian and non-Gaussian nonlinear filtering: Optimal and suboptimal approaches

Hamza Benzerrouk Polytechnique Montréal, Canada

Nonlinear state estimation is a fundamental challenge with many applications in areas such as target and multi-target tracking, navigation and control system design, power systems, biomedical engineering, econometrics, etc. In this talk, I will first present and compare state-of-the-art nonlinear filtering methods for Gaussian noise environments, such as extended Kalman filters, divided difference filters, unscented Kalman filters, cubature filters, Gauss-Hermite Kalman filters, etc., including their higher degree versions. For non-Gaussian noise environments, it is necessary to adapt and modify these filters to obtain more robust estimators. I will discuss and compare possible modifications and transformations based on the Gaussian sum approach, the point mass approach and the Monte Carlo approach, as well as Huber-based estimation.

Bio : Hamza Benzerrouk is a postdoctoral fellow at GERAD and in the department of Electrical Engineering at Polytechnique Montreal. He received his Ph.D. in Mathematics from Saint Petersburg State University, Russia, in 2014, after spending four years in the Department of Aerospace Instrumentation and Complex Information Computing. His fields of expertise include in Kalman filtering, nonlinear filtering, information and sensors fusion, radar target tracking, and integrated navigation and communication systems.


Coffee and biscuits will be offered at the beginning of the seminar.
Welcome to everyone!