Cubature Kalman Filters—CKF

MURI Meeting, UCSD, October 2008 (Haykin) 1

Cubature Kalman Filters

Simon Haykin

McMaster, University

Hamilton, Ontario, Canada

email: haykin@mcmaster.ca

Web site: http://soma.mcmaster.ca

MURI Meeting, UCSD, October 2008 (Haykin) 2

Outline of The Lecture

1. Introductory Remarks

MURI Meeting, UCSD, October 2008 (Haykin) 3

1. Introductory Remarks

In many applications, global optimality may not be practically feasible:

• Large-scale nature of the problem

• Infeasible computability

• Curse-of-dimensionality

Hence, the practical requirement of having to settle for a sub-optimal

solution of the system design

• Trade-off global optimality for computational tractability and robust

behaviour.

MURI Meeting, UCSD, October 2008 (Haykin) 4

Criterion for sub-optimality

MURI Meeting, UCSD, October 2008 (Haykin) 5

2. The Bayesian Filter: A powerful tool for

Problem statement:

• The Bayesian filter provides a unifying framework for the optimal

solution of this problem, at least in a conceptual sense.

Unfortunately, except in a few special cases, the Bayesian filter is not

implementable in practice -- hence the need for approximation.