*Tid:***18 september 2006 kl 16.15-17.00 **

*Plats :***Seminarierummet 3733**, Institutionen för
matematik, KTH, Lindstedts väg 25, plan 7. Karta!

*Föredragshållare:***Andrea Lang
**

* Titel: *
Estimation methods for terrain navigation
(Examensarbete)

* Sammanfattning: *
The subject of this thesis is terrain aided navigation: a navigation
technique that uses information about the surrounding terrain to
determine a vehicle's position. Integrated with an inertial
navigation system, this method yields fairly accurate position
estimates, which makes it a promising candidate for replacing the
satellite navigation system GPS in situations where the latter is
inconvenient or impossible to use.

The idea of terrain navigation is to compare measurements of the variations in the terrain height with a pre-stored digital map of the ground (or sea floor) topography, and from this estimate the position. The estimation procedure can be carried out in different ways. The aim of this thesis has been to implement two recursive position estimation methods, the particle filter and a method relying on the unscented transform, called unscented Kalman filter, and to compare them with each other as well as with two other methods implemented in an earlier work by Yingrong Xie: the point mass filter and a special variant of the ordinary Kalman filter. All of these methods rely on the Bayesian approach, in which the true position of the vehicle is regarded as a random variable, for which a probability density function can be modelled, instead of as a fixed but unknown parameter that is to be estimated. The performances of the four methods are also compared with the Cramér-Rao lower bound, which describes the theoretically best possible performance of an estimator given a certain filtering problem.