Tid: 27 mars2000 kl 1515-1600
Plats : Seminarierummet 3733, Institutionen för matematik, KTH, Lindstedts väg 25, plan 7. Karta!
Föredragshållare: Carl-Magnus Fahlcrantz
Titel: Computer-intensive methods in modern navigation. (Examensarbete)
Sammanfattning:
The rapid development of high-performing personal computers has lately led to a fast advancement in the field of computer-intensive methods. The introduction of the Global Positioning System (GPS) has given birth to a similar swift development in the field of integrated navigation systems, a combined use of different navigation sensors. Traditionally, linear methods such as the Kalman filter have been used to conduct the integration. The noise processes involved in the analysis are usually assumed to be white Gaussian noise, and the first part of this thesis deals with the true shape of such noise processes. Flight data from the New Integrated Navigation System, currently in developement at SAAB Gripen, was studied to analyse the real noise processes from this system. Due to the limited data available no definite result could be stated, but tendencies and suggestions for further analysis were made. The second part introduces non-linear alternatives to the extended Kalman filter. Results clearly indicate that the choice of integration model is of primary importance, and the use of non-linear filters might be an alternative to the Kalman filter in some situations. The results also clarify the weaknesses of non-linear filters under difficult circumstances such as high-dimensional environments, non-available natural noise processes, large measurement errors, and real-time applications.