Increasingly, for many application areas, it is becoming important to include elements of nonlinearity and non-Gaussianity in order to model accurately the underlying dynamics of a physical system. Moreover, it is typically crucial to process data on-line as it arrives, both from the point of view of storage costs as well as for rapid adaptation to changing signal characteristics. In this paper, we review both optimal and suboptimal Bayesian algorithms for nonlinear/non-Gaussian tracking problems, with a focus on particle filters. Particle filters are sequential Monte Carlo methods based on point mass (or "particle") representations of probability densities, which can be applied to any state-space model and which generalize the traditional Kalman filtering methods. Several variants of the particle filter such as SIR, ASIR, and RPF are introduced within a generic framework of the sequential importance sampling (SIS) algorithm. These are discussed and compared with the standard EKF through an illustrative example.

/pdf/a-tutorial-on-particle-filters-for-online-nonlinear-non-1rzkwgk2pp.pdf

A tutorial on particle filters for online nonlinear/non-Gaussian Bayesian tracking

Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

Today's Web-enabled deluge of electronic data calls for automated methods of data analysis. Machine learning provides these, developing methods that can automatically detect patterns in data and then use the uncovered patterns to predict future data. This textbook offers a comprehensive and self-contained introduction to the field of machine learning, based on a unified, probabilistic approach. The coverage combines breadth and depth, offering necessary background material on such topics as probability, optimization, and linear algebra as well as discussion of recent developments in the field, including conditional random fields, L1 regularization, and deep learning. The book is written in an informal, accessible style, complete with pseudo-code for the most important algorithms. All topics are copiously illustrated with color images and worked examples drawn from such application domains as biology, text processing, computer vision, and robotics. Rather than providing a cookbook of different heuristic methods, the book stresses a principled model-based approach, often using the language of graphical models to specify models in a concise and intuitive way. Almost all the models described have been implemented in a MATLAB software package--PMTK (probabilistic modeling toolkit)--that is freely available online. The book is suitable for upper-level undergraduates with an introductory-level college math background and beginning graduate students.

Machine Learning : A Probabilistic Perspective

Monte Carlo methods are revolutionizing the on-line analysis of data in fields as diverse as financial modeling, target tracking and computer vision. These methods, appearing under the names of bootstrap filters, condensation, optimal Monte Carlo filters, particle filters and survival of the fittest, have made it possible to solve numerically many complex, non-standard problems that were previously intractable. This book presents the first comprehensive treatment of these techniques, including convergence results and applications to tracking, guidance, automated target recognition, aircraft navigation, robot navigation, econometrics, financial modeling, neural networks, optimal control, optimal filtering, communications, reinforcement learning, signal enhancement, model averaging and selection, computer vision, semiconductor design, population biology, dynamic Bayesian networks, and time series analysis. This will be of great value to students, researchers and practitioners, who have some basic knowledge of probability. Arnaud Doucet received the Ph. D. degree from the University of Paris-XI Orsay in 1997. From 1998 to 2000, he conducted research at the Signal Processing Group of Cambridge University, UK. He is currently an assistant professor at the Department of Electrical Engineering of Melbourne University, Australia. His research interests include Bayesian statistics, dynamic models and Monte Carlo methods. Nando de Freitas obtained a Ph.D. degree in information engineering from Cambridge University in 1999. He is presently a research associate with the artificial intelligence group of the University of California at Berkeley. His main research interests are in Bayesian statistics and the application of on-line and batch Monte Carlo methods to machine learning. Neil Gordon obtained a Ph.D. in Statistics from Imperial College, University of London in 1993. He is with the Pattern and Information Processing group at the Defence Evaluation and Research Agency in the United Kingdom. His research interests are in time series, statistical data analysis, and pattern recognition with a particular emphasis on target tracking and missile guidance.

/pdf/sequential-monte-carlo-methods-in-practice-3kfxeebn8e.pdf

Sequential Monte Carlo methods in practice

Planning and navigation algorithms exploit statistics gleaned from uncertain, imperfect real-world environments to guide robots toward their goals and around obstacles.

Probabilistic Robotics

In this paper we consider the problem of tracking a maneuvering target in clutter. We apply an original on-line Monte Carlo filtering algorithm to perform optimal state estimation. Improved performance of the resulting algorithm over standard IMM/PDAF based filters is demonstrated.© (1999) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

Sequential Monte Carlo for maneuvering target tracking in clutter

We have previously discussed the potential of using an Hg1-xCdxTe (MCT) source as a reference plane for the nonuniformity
correction of thermal imagers. Due to the fast switching speed, the apparent temperature can be changed on a
frame to frame basis. This allows multipoint correction data to be obtained without having to wait for temperatures to
stabilize as with a Peltier reference source. Also, the operation of the device can be synchronized to the integration
period of the camera to reduce the mean power requirements by the ratio of the frame to the integration time and hence
thermal heating effects are also reduced.
In this paper, we discuss a practical implementation of this concept in a thermal imaging camera, which is being
developed as part of the UK MOD Albion program. This development has involved increasing the device size,
increasing the effective temperature range and matching the drive requirements to typical camera power supplies. The
factors determining the achievable effective temperatures are discussed, together with modifications to the device design
that have been implemented to obtain a useful temperature range. The drive requirements have been improved by
developing a series connected structure. This has reduced the peak current by a factor of 4 and allows the devices to be
controlled with conventional Peltier reference electronics rather than a custom unit. The improved devices have now
been incorporated into a state-of-the-art infrared camera and their performance in this system will be discussed.

Improved MWIR reference sources for FPA non-uniformity correction

Orbit Determination Analysis for a Joint UK-Australian Space Surveillance Experiment

We have previously presented results from our mercury cadmium telluride (MCT, Hg1-xCdxTe) growth on silicon
substrate technology for different applications, including negative luminescence, long waveband and mid/long dual
waveband infrared imaging. In this paper, we review recent developments in QinetiQ's combined molecular beam
epitaxy (MBE) and metal-organic vapor phase epitaxy (MOVPE) MCT growth on silicon; including MCT defect
density, uniformity and reproducibility. We also present a new small-format (128 x 128) focal plane array (FPA) for high
frame-rate applications.
A custom high-speed readout integrated circuit (ROIC) was developed with a large pitch and large charge storage aimed
at producing a very high performance FPA (NETD ~10mK) operating at frame rates up to 2kHz for the full array. The
array design allows random addressing and this allows the maximum frame rate to be increased as the window size is
reduced. A broadband (2.5-10.5 μm) MCT heterostructure was designed and grown by the MBE/MOVPE technique onto
silicon substrates. FPAs were fabricated using our standard techniques; wet-etched mesa diodes passivated with epitaxial
CdTe and flip-chip bonded to the ROIC.
The resulting focal plane arrays were characterized at the maximum frame rate and shown to have the high operabilities
and low NETD values characteristic of our LWIR MCT on silicon technology.

Wide-band (2.5 - 10.5 Âµm), high-frame rate IRFPAs based on high-operability MCT on silicon

One of the main advantages of increasing the operating temperature of infrared focal plane arrays (FPAs) is to take
advantage of lower cost cooling options such as thermoelectric coolers. However the maximum reduction in temperature
available from the current generation of coolers (e.g. 4-stage) is around 110 K. For a maximum operating temperature of
70 oC, this means that the FPA needs to operate above 233 K. In this region, the performance becomes a strong function
of array temperature and designing a system becomes a trade-off between the performance of the fpa; the speed of the
optics; the maximum temperature of operation; and the cooler power and complexity.
In this paper, previous results will be extrapolated to estimate the FPA performance across this trade space by varying
cut-off wavelength. Possible techniques to enhance the performance of the FPAs by reducing low frequency noise or
adding optical concentrators will also be considered. These extrapolated results indicate that in an f/2 system at 210 K,
an NETD of around 30 mK could be achievable. Potential applications for the technology are in systems where long
lifetime; no moving parts; or reduced weight are an advantage. Ideally the maximum ambient temperature should be
limited to maintain the best thermal sensitivity. Suitable applications could include sensors which operate from UAVs or
in space.

Neil Gordon

Papers

Sequential Monte Carlo for maneuvering target tracking in clutter

Improved MWIR reference sources for FPA non-uniformity correction

Orbit Determination Analysis for a Joint UK-Australian Space Surveillance Experiment

Wide-band (2.5 - 10.5 Âµm), high-frame rate IRFPAs based on high-operability MCT on silicon

Thermo electrically cooled focal plane arrays based on MCT