Particle Filters for Random Set Models

ISBN-13: 9781489988843 / Angielski / Miękka / 2015 / 174 str.

Branko Ristic

Opis

This bookdiscusses state estimation of stochastic dynamic systems from noisy measurements, specifically sequential Bayesian estimation and nonlinear or stochastic filtering. The class of solutions presented in this book is based on the Monte Carlo statistical method. Although the resulting algorithms, known as particle filters, have been around for more than a decade, the recent theoretical developments of sequential Bayesian estimation in the framework of random set theory have provided new opportunities which are not widely known and are covered in this book. This book is ideal for graduate students, researchers, scientists and engineers interested in Bayesian estimation.

Recenzje

From the book reviews:

"The book realizes a happy union between theory and practice. Of high interest are the Algorithms for which their pseudo-codes are presented. We think we are faced with an excellent book that will have a great success and audience between those interested for new approaches in filtering theory." (Dumitru Stanomir, zbMATH 1306.93002, 2015)

Spis treści

3.3.2 Classification results References 4 Multi-object particle filters 4.1 Bernoulli particle filters 4.1.1 Standard Bernoulli particle filters 4.1.2 Bernoulli box-particle filter 4.2 PHD/CPDH particle filters with adaptive birth intensity 4.2.1 Extension of the PHD filter 4.2.2 Extension of the CPHD filter 4.2.3 Implementation

4.2.4 A numerical study 4.2.5 State estimation from PHD/CPHD particle filters 4.3 Particle filter approximation of the exact multi-object filter References 5 Sensor control for random set based particle filters 5.1 Bernoulli particle filter with sensor control 5.1.1 The reward function 5.1.2 Bearings only tracking in clutter with observer control 5.1.3 Target Tracking via Multi-Static Doppler Shifts 5.2 Sensor control for PHD/CPHD particle filters 5.2.1 The reward function 5.2.2 A numerical study 5.3 Sensor control for the multi-target state particle filter 5.3.1 Particle approximation of the reward function 5.3.2 A numerical study References 6 Multi-target tracking 6.1 OSPA-T: A performance metric for multi-target tracking 6.1.1 The problem and its conceptual solution 6.1.2 The base distance and labeling of estimated tracks 6.1.3 Numerical examples 6.2 Trackers based on random set filters 6.2.1 Multi-target trackers based on the Bernoulli PF 6.2.2 Multi-target trackers based on the PHD particle filter 6.2.3 Error performance comparison using the OSPA-T error 6.3 Application: Pedestrian tracking 6.3.1 Video dataset and detections 6.3.2 Description of Algorithms 6.3.3 Numerical results References 7 Advanced topics 7.1 Bernoulli filter for extended target tracking 7.1.1 Mathematical models 7.1.2 Equations of the Bernoulli filter for an extended target 7.1.3 Numerical Implementation 7.1.4 Simulation results 7.1.5 Application to a surveillance video 7.2 Calibration of tracking systems 7.2.1 Background and problem formulation 7.2.2 The proposed calibration algorithm 7.2.3 Importance sampling with progressive correction 7.2.4 Application to sensor bias estimation References Index

O autorze

Branko Ristic is at the Defence Science and Technology Organisation, Australia

Defence Science and Technology Organisation, Australia

Blurb

“Particle Filters for Random Set Models” presents coverage of state estimation of stochastic dynamic systems from noisy measurements, specifically sequential Bayesian estimation and nonlinear or stochastic filtering. The class of solutions presented in this book is based  on the Monte Carlo statistical method. The resulting  algorithms, known as particle filters, in the last decade have become one of the essential tools for stochastic filtering, with applications ranging from  navigation and autonomous vehicles to bio-informatics and finance.

While particle filters have been around for more than a decade, the recent theoretical developments of sequential Bayesian estimation in the framework of random set theory have provided new opportunities which are not widely known and are covered in this book. These recent developments have dramatically widened the scope of applications, from single to multiple appearing/disappearing objects, from precise to imprecise measurements and measurement models.

This book is ideal for graduate students, researchers, scientists and engineers interested in Bayesian estimation.