4.0
4.0 | 1 rating Rate this file 21 Downloads (last 30 days) File Size: 2.94 KB File ID: #36301 Version: 1.0
image thumbnail

Extended Kalman Filter Tracking Object in 3-D

by

Alex Dytso (view profile)

 

Using Kalman filter to track object in 3D. Comparing Extended Kalman filter to its linear version.

| Watch this File

File Information
Description

Assume that we want to track an object moving in 3-D space with constant velocity. Our instruments observe bearing, range and high(cylindrical coordinates). However, of an interest are rectangular coordinates. Since transformation is non-linear this requires use of extended Kalman filter.
Because transformation is non-linear between X,Y and Range,Bearing and linear between Z and high(Z is height), this problems serves as a good comparason of how well extended Kalman filter can perform. By comparing its linear estimation error in Z to non-linear estimations in X and Y, we can judge how non-familiarities effect estimation.

MATLAB release MATLAB 7.6 (R2008a)
Tags for This File   Please login to tag files.
Please login to add a comment or rating.
Comments and Ratings (11)
14 May 2014 Ahsan

Ahsan (view profile)

I give the X(:,1)=MAT(1,:)' as actual initial condition, where MAT is the matrix of [501x6] and i'm confusing about initial observation `Z` and assumed initial condition `Xh`

Comment only
14 May 2014 Ahsan

Ahsan (view profile)

The value of Z is unused from argument in proccesANDobserve and Jacobian function.

Comment only
14 May 2014 Ahsan

Ahsan (view profile)

I know this is the observation vector, I edited a bit of your code for my purpose, but it crosses the actual trajectory and calculating in its opposite way. I have a matrix `MAT` of [501x6] having 1:3 for position and 4:6 for velocities, How can I set the initial observation vector and also what other initial assumptions would be set?

Comment only
07 May 2014 Alex Dytso

Alex Dytso (view profile)

'Z' Stands for the observation vector and it is used in number of places for example when you compute quantity called innovation.

Comment only
07 May 2014 Ahsan

Ahsan (view profile)

Hello, I didn't understand the Use of `Z` as this is unused in your code. Its always calculating but didn't use the initial array.

Comment only
30 Sep 2013 yatie SUAIB

ok,thank you very much

Comment only
26 Sep 2013 Alex Dytso

Alex Dytso (view profile)

Yes, here is the document this is based on
https://dl.dropboxusercontent.com/u/12025879/Extended%20Kalman%20Filter.pdf

Comment only
26 Sep 2013 yatie SUAIB

Hi Alex
Do you have the article/journal paper that you are referring to in order to write these Matlab codes?

Comment only
31 Mar 2013 Alex Dytso

Alex Dytso (view profile)

In order to convert to 2-D you just have to change the appropriate dimensions of matrices. You can also use the code as is and ignore one of the outputs.

Comment only
31 Mar 2013 Atef

Atef (view profile)

please how apply this code for 2-D ?

Comment only
30 Dec 2012 W. Chong  

Contact us