A Brief Tutorial On Recursive Estimation With Examples From Intelligent Vehicle Applications (Part IV): Sampling Based Methods And The Particle Filter

Following the third article of the series "A brief tutorial on recursive estimation", in this article (the fourth article) we continue to focus on the problem of how to handle model nonlinearity in recursive estimation. We will review the particle filter a.k.a. a sequential Monte Carlo method which has the potential to handle recursive estimation problems with an system model and a measurement model of arbitrary types and with data statistics of arbitrary types. We will explain basic principles that underlie the particle filter, and demonstrate its performance with examples from intelligent vehicle applications. We will explain its advantage as well as limitation.