Skip to content

Latest commit

 

History

History
60 lines (41 loc) · 3.69 KB

README.md

File metadata and controls

60 lines (41 loc) · 3.69 KB
Raw

Linear Kalman Filter Algorithm

This is a C++ implementation of the Linear Kalman Filter, designed for use with the ROS Framework. Even though the code has been written for a specific use case, it can be used anywhere with a few minor changes. It uses the Eigen library for the matrix math, so make sure you have that installed. See : Getting started with Eigen

Please read Bzarg's website to know the basics of the working of the Kalman Filter algorithm.It is an excellent website and the working of the Kalman Filter has been explained pretty nicely. All the variable names used are taken from this website.

To use this filter, make a new workspace and clone the files into the src folder. Make sure you are in the root directory

mkdir -p ~/Filter/src
cd ~/Filter/
catkin build
cd src
git clone https://github.com/Kshitij1K/Linear-Kalman-Filter/

Now the filter has to be modified for your use case. You need to make changes to the CMakeLists.txt, Params.yaml, Main.cpp, Prediction.h, Update.h, Processing.h files.

Changes to the CMakeLists.txt

In the following code (Line 4)

find_package(catkin REQUIRED COMPONENTS
geometry_msgs
roscpp
sensor_msgs
std_msgs
cmake_modules


)

add the dependencies required by your sensors.

Changes to Params.yaml

You need to change every parameter here, according to your sensor model. All the parameters are in the form of matrices, however they have been written linearly in the YAML file. Detailed instructions are commented in the YAML file.

Changes to Main.cpp

In the line 17, change the message type inside the angular brackets <> to whatever message type you want to publish.

Changes to Prediction.h

In the line 6, change the argument of the function to whatever message type you are going to use for the prediction step. It should be a constant pointer.

Changes to Update.h

In the line 6, change the argument of the function to whatever message type you are going to use for the update step. It should be a constant pointer.

Changes to Processing.h

In this file, you need to do the following

  1. Add the necessary includes for the messages used in prediction, measurement, and update steps.(After line 11)
  2. Declare the output message, by changing it to whatever message type you want to publish (line 16)
  3. Change the functions UkProcess, ZkProcess and OutputProcess functions. These functions basically take the data in sensor messages, put them in matrix form, and return the matrices(UkProcess and ZkProcess) and vice-versa(OutputProcess). In case any processing on the messages is required, add them in the corresponding functions. (I have done some processing in the UkProcess and OutputProcess functions, remove those if necessary)
Changes to KF.launch

In line 6,7,8 change the remappings of Measurement_data, Prediction_data and Filtered_data to the corresponding topics which you are going to subscribe/publish. Measurement_data should be remapped to the topic on which data for the measurement step is going to be recieved, Prediction_data should be remapped to the topic on which data for the prediction step is going to be recieved and Filtered_data should be remapped to the topic on which you are going to publish the filtered data

After all these changes have been made, you can build and launch the filter using the following commands. Make sure you are in the root directory

cd Filter
catkin build
source devel/setup.bash
roslaunch filter KF.launch

And you are good to go! If you have any problems, please open a new issue on this repository