Need help: Understanding Kalman Filter (measurement data)

Hello, I'm trying to understand the Kalman Filter through this example: http://www.mathworks.com/matlabcentral/fileexchange/31977-kalman-filter-example/content/myKalman.html

My question is, why if I change zHat_t = C*x_t + Ez; into zHat_t = 0; there is no effect into the results of the estimation. Is it mean that the measurement data don't effect the estimation results?

Answer 1

[Youssef Khmou]

hi,

NO that equation depends on the system on which you are applying Kalman filter . you can start first by looking at this famous tutorial which explains the theory first before trying to apply the example .

http://www4.cs.umanitoba.ca/~jacky/Teaching/Courses/74.795-LocalVision/ReadingList/SIGGRAPH2001_CoursePack_08.pdf

take a look at pages 13-16

[K]

Thanks for your reply. So regardless of any measurement data (in this case the measured position), the estimation results will only depends on the system? So the measurement data is not important?

[Youssef Khmou]

no it is important, suppose you are estimating the velocity of vehicule but what you are measuring is instantaneous position in discrete time so : V(n)= H *x(n) +W(n) W~N(m,var) so H would be the derivation operator to obtain the velocity, however when the data is measured directly like a temperature for example, and the system is noiseless then the second equation is not important, its then the same as : X(n+1)=AX(n)+V(n) ,