Skip to content

Latest commit

 

History

History
85 lines (61 loc) · 2.71 KB

README.md

File metadata and controls

85 lines (61 loc) · 2.71 KB
Raw

Bayesian Filtering Library

This project provides a robust implementation of various Bayesian filtering techniques, including Linear Kalman Filter, Extended Kalman Filter, Unscented Kalman Filter, and Particle Filter. It's designed for personal use for robotics applications.

Features

  • Linear Kalman Filter: Implements the basic Kalman filter for linear models.
  • Extended Kalman Filter: Supports non-linear models by linearizing about the current estimate.
  • Unscented Kalman Filter: Uses the unscented transform to handle highly non-linear models better than EKF.
  • Particle Filter: Implements a sequential Monte Carlo method for non-linear and non-Gaussian problems.
  • Data Generation: Tools to generate synthetic data for testing and validation.
  • Visualization: Integrates with modern visualization tools to provide real-time insights into filter performance.

Getting Started

Prerequisites

  • C++ Compiler (g++ 9.4.0)
  • CMake (version 3.10 or higher)
  • Eigen3 Library
  • Matplotlib C++

Building the Project

To build the project, follow these steps:

mkdir build
cd build
cmake ..
make

This will compile all targets defined in the project including the library, examples, and tests.

Running Examples

After building, you can run the examples:

./build/examples/demo

Usage

Here is a simple example on how to use the Linear Kalman Filter:

#include <Eigen/Dense>
#include "kf/LinearKalmanFilter.hpp"

int main() {
    Eigen::MatrixXd transition(2, 2);
    transition << 1, 1, 0, 1;
    Eigen::MatrixXd observation(1, 2);
    observation << 1, 0;
    Eigen::MatrixXd processNoise(2, 2);
    processNoise << 0.1, 0, 0, 0.1;
    Eigen::MatrixXd measurementNoise(1, 1);
    measurementNoise << 0.1;
    LinearKalmanFilter kf(transition, observation, processNoise, measurementNoise);
    
    Eigen::VectorXd observation(1);
    observation << 1; // Example observation data

    kf.predict();
    kf.update(observation);

    return 0;
}

Contributing

Contributions to enhance the functionality or efficiency of this filtering library are welcome. Please feel free to fork the repository, make changes, and submit a pull request.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Authors

  • Yunus Talha Erzurumlu - Primary Developer - yunusstalha

Acknowledgments

  • Thanks Dr. Umut Orguner and Dr. Emre Ozkan, for their lectures notes to make this topic clear for me.

References

  • For a detailed introduction to Kalman filters and Bayesian filtering techniques, see the EE793 lecture notes by Umut Orguner available here.