Object Tracking over Sensor Networks

Lead Research Organisation: Lancaster University
Department Name: Communications Systems

Abstract

Object tracking is an important problem that has been of interest to researchers from different fields. Tracking algorithms are used in a wide variety of domains, such as robotics, vehicular traffic, navigation and communication systems. The main goal is to obtain a record of the trajectory of the moving object(s) over space and time by processing the sensor data. Reliable tracking methods are of crucial importance in many surveillance systems in order to enable human operators to remotely monitor activity across large environments such as: a) transport systems (e.g., railway transportation, airports, urban and motorway road networks, and maritime transportation), b) banks, shopping malls, car parks, and public buildings, c) industrial environments, and d) government establishments (military bases, prisons, strategic infrastructures, radar centres, and hospitals). The problem has different particularities depending on whether data from one or multiple sensors are used. Sensor networks offer many advantages due to the fact that the coming data provide a global picture from different sides. Multiple-sensor systems can provide surveillance coverage across a wide area, ensuring constant object visibility. However, the presence of data from multiple sensors poses many new challenges from theoretical and practical point of view that will be addressed in this project. How to efficiently process the data from many sensors is a significant problem. The data rates associated with collection assets can vary greatly, for instance between a measurement once from each day from a satellite to a measurement every twenty-fifth of a second from a camera. It is impossible to define a single tracking methodology and technique that meet the requirements of all these domains. Advanced tracking algorithms can combine the functionality of existing identification and tracking processes while accounting for any uncertainty if present.Different techniques will be developed in this project outperforming the previously existing techniques in the literature, which will be suitable for on-line implementations. The main interest will be focussed on innovative Bayesian techniques, such as sequential Monte Carlo methods (also called particle filters), Monte Carlo Markov chains and Unscented Kalman filtering, providing efficient approximations to the optimal Bayesian solutions. The Monte Carlo approach is generic, scalable, flexible and has opportunities for parallelisation and distributed implementation. Monte Carlo methods afford natural incorporation of constraints which is difficult or impossible with standard filtering techniques. The algorithms will be implemented in a centralised and distributed way, which is a novel and significant achievement. It can save a lot of energy and reduce the communication load of the supporting sensor networking systems, it will increase its robustness to failure and respectively the reliability of the tracking module. The innovative elements of this proposal rely on the powerful methodology and the focus on very important problems that have been in the scope of interest of scientists and engineers. Problems such as group object tracking and distributed particle filtering represent substantial research challenges which makes this research unique.

Publications

10 25 50

publication icon
A. Gning (2011) A box particle filter for stochastic and set-theoretic measurements with association uncertainty in International Conference on Information Fusion

publication icon
A. Gning, L. Mihaylova, F. Abdallah, B. Ristic (2012) Integrated Tracking, Classification and Sensor Management. Theory and Applications

publication icon
Angelova D (2008) Extended Object Tracking Using Monte Carlo Methods in IEEE Transactions on Signal Processing

publication icon
Gning A (2011) Interval Macroscopic Models for Traffic Networks in IEEE Transactions on Intelligent Transportation Systems

publication icon
Gning A. (2009) Dynamic clustering and belief propagation for distributed inference in random sensor networks with deficient links in 2009 12th International Conference on Information Fusion, FUSION 2009

publication icon
Mihaylova L (2011) Localization of Mobile Nodes in Wireless Networks with Correlated in Time Measurement Noise in IEEE Transactions on Mobile Computing

publication icon
Mihaylova L (2012) Parallelized Particle and Gaussian Sum Particle Filters for Large-Scale Freeway Traffic Systems in IEEE Transactions on Intelligent Transportation Systems

publication icon
Mihaylova L (2007) Mobility Tracking in Cellular Networks Using Particle Filtering in IEEE Transactions on Wireless Communications

publication icon
Mihaylova L. (2009) Parallelised gaussian mixture filtering for vehicular traffic flow estimation in INFORMATIK 2009 - Im Focus das Leben, Beitrage der 39. Jahrestagung der Gesellschaft fur Informatik e.V. (GI)

 
Description Main achievements
1. Group and extended object tracking

We proposed algorithms for group object structure and motion estimation based on random evolving networks combined with sequential Monte Carlo algorithms. The main contributions of our work for group object tracking consist in: i) the developed graphical representation of the groups, for different types of motion, including interactions and taking into account the measurement origin uncertainty, ii) target state estimation with proposed sequential Monte Carlo methods.

Evolutionary graph network-type models for the group structure are proposed. The graph structure can be deterministically estimated or in a probabilistic way with a graph jointly updated with the samples of the particle filter. Then a particle of the developed sequential Monte Carlo filters contains the graph structure of the group which gives significant information. In this graph the connected components correspond to groups of targets. The effectiveness of the proposed techniques is investigated and validated over a challenging urban environment scenario with splitting, merging and crossing of groups. The performance of the approach is also validated over real ground moving target indicator data sets. The proposed approaches successfully estimate the targets states and the group structure graph with reliable performance and accurate tracking.

Distributed Algorithms and Optimisation of the Resources of Sensor Networks
A fundamental issue in real-world monitoring sensor network systems is the choice of sensors to track local events. Ideally, the sensors work together, in a distributed manner, to achieve a common mission-specific task. We developed a framework for distributed inference based on dynamic clustering and belief propagation in sensor networks with deficient links. We investigated this approach for dynamic clustering of sensor nodes combined with belief propagation for the purposes of object tracking in sensor networks with and without deficient links. The efficiency of our approach is demonstrated over an example of hundreds randomly deployed sensors. Optimisation of the resources of the sensor network was considered, for providing optimal use of the number of sensors, while achieving desirable estimation accuracy and complexity suitable for real time. Dynamic clustering is combined jointly with belief propagation for the purposes of object tracking.

3. Parallelised/ Distributed Particle Filters for Vehicular Traffic Flow Estimation
Large traffic network systems require handling huge amounts of data, often distributed over a large geographical region in space and time. Centralised processing is not then the right choice in such cases. Distributed/ parallelised techniques are needed for processing large amounts of data.

We developed in a parallelised Gaussian Mixture Model filter (GMMF) for vehicular traffic networks aimed to: 1) work with high amounts of heterogenous data (from different sensor modalities), 2) provide robustness in the presence of sparse and missing sensor data, 3) able to incorporate different models in different traffic segments and represent various traffic regimes, 4) able to cope with multimodalities (e.g., due to multimodal measurement likelihood or multimodal state probability density functions). The efficiency of the parallelised GMMF is investigated over traffic flows based on macroscopic modelling and compared with a centralised GMMF. Parallelised Gaussian Sum Particle filters are developed and compared with parallelised particle filters (with efficient proposal functions). The proposed GMM approach is general, it is applicable to systems where the overall state vector can be partitioned into state components (subsets), corresponding to certain geographical regions, such that most of the interactions take place within the subsets.

4. Localisation and Mobility Tracking in Wireless Sensor Networks
Within this EPSRC project the problem of localisation in wireless networks (cellular and ad hoc) has been studied and several sequential Monte Carlo techniques are developed: a multiple model particle filter, Rao-Blackweliised and auxiliary particle filters (both for correlated in time measurements and measurements with unknown noise characteristics).

5. Box Particle Filtering
A new methodology called Box Particle Filtering (BPF) is also theoretically studied and validated over simulated and real data. The BPF is a very promising methodology for complex and high dimensional systems. Box particle filters afford achieving the same estimation accuracy with dozens of particles, whereas generic particle filters needs thousands to achieve the same accuracy.
Exploitation Route The theoretical methods are published in journal and conference papers.
The software for the key developed algorithms is made available as open-source, for everyone to use and can be downloaded from the Mathworks Central web sites:

Matlab Central code for the Box Particle Filter and Bernoulli Box Particle Filter,

* http://www.mathworks.com/matlabcentral/fileexchange/43012-box-particlefilter-
and-bernoulli-box-particle-filter, 2013.
* Matlab Central code for group object structure and state estimation with
evolving networks and Monte Carlo methods, http://www.mathworks.com/
matlabcentral/fileexchange/43906, 15 Oct. 2013.
* Matlab Central code for the Gaussian mixture particle algorithm for dynamic
cluster tracking, http://www.mathworks.com/matlabcentral/fileexchange/
44298-the-gaussian-mixture-particle-algorithm-for-dynamic-cluster-tracking,
14 November 2013.
* Matlab code for "Mobility Tracking in Cellular
Networks Using Particle Filtering", http://www.mathworks.com/matlabcentral/fileexchange/47428-pf-programs-zip
Sectors Aerospace, Defence and Marine,Digital/Communication/Information Technologies (including Software),Healthcare,Transport

 
Description The software and algorithms are used in the areas of surveillance and security.
First Year Of Impact 2010
Sector Digital/Communication/Information Technologies (including Software)
Impact Types Societal

 
Description Data Information Fusion Defence Technology Centre (DIF DTC)
Amount £250,000 (GBP)
Organisation Ministry of Defence (MOD) 
Sector Public
Country United Kingdom
Start 10/2007 
End 07/2009
 
Description FP7 Marie Curie Initial Training Networks - MC IMPULSE project, http://mcimpulse.isy.liu.se/
Amount € 400,000 (EUR)
Funding ID No. 238710 (Monte Carlo based Innovative Management and Processing for an Unrivalled Leap in Sensor Exploitation, MC IMPULSE) 
Organisation European Commission 
Sector Public
Country European Union (EU)
Start 10/2009 
End 10/2014
 
Description Fraunhofer Institute (FKIE): collaboration via two ITN Marie Curie projects, MC IMPULSE and TRAX 
Organisation Fraunhofer Society
Department Fraunhofer Institute FKIE
Country Germany 
Sector Academic/University 
PI Contribution Joint research collaboration in the areas of sensor data fusion and multiple target tracking. Web link to the current EU ITN Marie Curie project, TRAX: http://cordis.europa.eu/project/rcn/109279_en.html The web link to the completed EU ITN Marie Curie project, called MC IMPULSE: https://mcimpulse.isy.liu.se/
Collaborator Contribution Developments of methods for sensor data fusion and multiple target tracking.
Impact Joint publications as listed below: * M. Schikora, A. Gning, L. Mihaylova, D. Cremers, W. Koch, Box-Particle Hypothesis Density Filter for Multi-Target Tracking, IEEE Transactions on Aerospace and Electronic Systems, Vol. 50, No. 3, July, 2014, DOI:10.1109/TAES.2014.120238, in print. * N. Petrov, M. Ulmke, L. Mihaylova, A. Gning, M. Schikora, M. Wieneke and W. Koch, On the Performance of the Box Particle Filter for Extended Object Tracking Using Laser Data, Proc. from the IEEE Sensor Data Fusion Workshop: Trends, Solutions, Applications, Bonn, Germany, 4-6 Sept. 2012, pp. 19 - 24.
Start Year 2008
 
Description Linkoeping University, Sweden 
Organisation Linkoping University
Country Sweden 
Sector Academic/University 
PI Contribution Joint research in the area of signal processing and novel algorithms for filtering and decision making.
Collaborator Contribution Knowledge exchange
Impact The outcomes from this collaboration related with the two ITN Marie Curie projects (MC IMPULSE that funished) and the current ITN Marie Curie project, called TRAX are in multidisciplinary areas - engineering, ICT, applied mathematics and have applications to intelligent transportation systems, surveillance and other areas.
Start Year 2009
 
Description QinetiQ 
Organisation Qinetiq
Country United Kingdom 
Sector Private 
PI Contribution Results in the areas of sensor data fusion, detection, tracking, reasoning over sensor networks and scene understanding.
Collaborator Contribution Knowledge exchange, real testing examples and data
Impact Publications * H. Bhaskar, L. Mihaylova, S. Maskell, Automatic Human Body Parts Detection Based on Cluster Background Subtraction and Foreground Learning, Neurocomputing. Special Issue on Behaviours in Video, Vol. 100, No. 1, pp. 58-72, 2013. * A. Gning, L. Mihaylova, S. Maskell, S. K. Pang, S. Godsill, Ground Target Group Structure and State Estimation with Particle Filtering, Proc. of the 11th International Conf. on Information Fusion, Cologne, Germany, pp. 1176 -- 1183, 30 June - 3 July 2008. * A. Gning, L. Mihaylova, S. Maskell, S. K. Pang, S. Godsill, Group Object Structure and State Estimation with Evolving Networks and Monte Carlo Methods, IEEE Transactions on Signal Processing, Vol. 59, No. 4, April, 1383 -- 1396, 2011.
Start Year 2006
 
Description Thales (NL) via two EU Initial Training Network Projects, MC Impulse, and now a second project, called TRAX 
Organisation Thales Group
Country France 
Sector Private 
PI Contribution We developed novel algorithms for nonlinear estimation, with various applications such as intelligent transportation systems and surveillance.
Collaborator Contribution Provision of data, practical testing examples and scenarios, knowledge exchange. Web link to the current EU ITN Marie Curie project, TRAX (Training network on tRAcking in compleX sensor systems): http://cordis.europa.eu/project/rcn/109279_en.html Web link to the completed EU ITN Marie Curie project, MC IMPULSE https://mcimpulse.isy.liu.se/
Impact This is a multi-disciplinary collaboration, in the areas of engineering, mathematics, signal processing and ICT.
Start Year 2009
 
Description University of Cambridge 
Organisation University of Cambridge
Country United Kingdom 
Sector Academic/University 
PI Contribution Development of novel sequential Monte Carlo filtering approaches. Joint publications: * L. Mihaylova, A. Carmi, F. Septier, A. Gning, S. K. Pang, S. Godsill, Overview of Sequential Bayesian Monte Carlo Methods for Group and Extended Object Tracking, Digital Signal Processing, February, 2014, Vol. 25, pp. 1-16. * A. Y. Carmi, L. Mihaylova, S. J. Godsill (Editors), Compressed Sensing and Sparse Filtering}, Series Signals and Communication Technology, Springer, Berlin Heidelberg, ISBN 978-3-642-38397-7, 2014. * A. Carmi, L. Mihaylova, A. Gning, P. Gurfil, S. Godsill, Inferring Leadership from Group Dynamics Using Markov Chain Monte Carlo Methods, Chapter in Modeling, Simulation, and Visual Analysis of Crowds}, Eds. S. Ali, K. Nishino, D. Monacha, M. Sha, Springer, pp. 335-357, 2014. * A. Carmi, L. Mihaylova, A. Gning, P. Gurfil, S. Godsill, Markov Chain Monte Carlo Based Autonomous Tracking and Causality Reasoning, Chapter in Advances in Intelligent Signal Processing and Data Mining: Theory and Applications, Eds. P. Georgieva, L. Mihaylova and L. Jain, pp. 7-53, 2013. * A. Gning, L. Mihaylova, S. Maskell, S. K. Pang, S. Godsill, Group Object Structure and State Estimation with Evolving Networks and Monte Carlo Methods, IEEE Transactions on Signal Processing}, Vol. 59, No. 4, April, 1383 - 1396, 2011. * A. Carmi, L. Mihaylova, F. Septier, S. Pang, S. Godsill, MCMC-Based Tracking and Identification of Leaders in Groups, Proceedings of the First IEEE Workshop on Modeling, Simulation and Visual Analysis of Large Crowds, Barcelona, Spain, 6 - 13 November 2011. * A. Gning, L. Mihaylova, S. Maskell, S. K. Pang, S. Godsill, Ground Target Group Structure and State Estimation with Particle Filtering, Proc. of the 11th International Conf. on Information Fusion}, Cologne, Germany, pp. 1176 - 1183, 30 June - 3 July 2008. * A. Gning, L. Mihaylova, S. Maskell, S. K. Pang, S. Godsill, Evolving Networks for Group Object Motion Estimation, Proc. of the Institution of Engineering and Technology (IET) Seminar on Target Tracking and Data Fusion: Algorithms and Applications}, 15-16 April 2008, Birmingham, UK, pp. 99-106.
Collaborator Contribution Joint development of methods for nonlinear filtering, including knowledge exchange.
Impact The outcomes are in multidisciplinary areas of engineering, ICT and mathematics.
Start Year 2007