Aussi en

State estimation by interval analysis and probabilistic approaches:
Box particle filtering and localisation applications

Logo from ENSTA Bretagne Logo de l'université d'Angers Logo de l'ISTIA IRI Logo

Info

This is the website of the project carried out by Evandro Bernardes from 14th of April 2017 to 29th of September of 2017 during his work placement project at IRI, in Barcelona.

This project is serves both as Evandro Bernardes' final graduation project at ENSTA Bretagne and as the final project for the Master of Research at ISTIA, from the University of Angers.

During the project, the Box Particle Filter algorithm, as described in the article State estimation and fault detection using box particle filtering with stochastic measurements (Blesa et al. 2015).

This project has been done under the supervision of Joaquim Blesa Izquierdo

Abstract

In this project, a box particle filtering algorithm is proposed and implemented. The algorithm combines two types of noise in measurements: stochastic noise affecting the measurements and bounded noise. The method is explained and compared extensively with the conventional particle filtering algorithm, both in a theoretical and in a practical way.

Multiple different simulations are performed to further analyse the differences in performance between the conventional particle filter and the box particle filter, including a simple SLAM application.

Keywords: State estimation. Interval analysis. Particle filter, box particle filter, SLAM.