Skip to content
No description or website provided.
Python Shell
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.

Monte Carlo Particle Filter for Localization

Particle Filter Algorithm is a nonparametric implementation of the Bayes Filter to approximate state, for example of a robot moving in a maze.

The idea is to represent the posterior belief by a finite number of random variables (particles). The algorithm is repeatedly resampling those particles based on likelihood derived from a measurement model.

This project is a examination project for SS18 Monte Carlo Methods in Artificial Intelligence and Machine Learning course taught by Prof. Dr. Manfred Opper and Theo Galy-Fajou, at TU Berlin.


pip install -r requirements.txt


> python

    - scene :  [scene-1, scene-2, scene-1-kidnapping, scene-2-kidnapping, scene-8.12]
    - no_particles : number of particles, for example 100.
    - total_frames :  total time step to run the simulation, it's useful for debugging.
    - show_particles : a boolean option whether to show the particles.
    - no_random_particles: number of random particles introduced to the system, required for kidnapping scenes.
    - save : a boolean option whether to see the simulartor live or save the result as a video.
    - frame_interval : time interval for each frame, default is 50s.


python --scene scene-1 --no-particles 100 --save


Anders Dahl Hjort, Luis Dreisbach, and Pattarawat Chormai

You can’t perform that action at this time.