You are not logged in. Click here to Log in

Fellow Projects

There are 12 projects for Early Stage Researchers (ESR, 3 year projects) and 3 projects for Experienced Researchers (ER, 2 year projects).



ESR Projects

Interpolation and model estimation aspects of large scale spatial data


Host:Linköping university
Supervisor:GUSTAFSSON Fredrik
Description:The development of new algorithms and models for spatial interpolation or radio measurements indoors. Bayesian approaches will be adopted to fuse measurement data uncertainty with measurement position uncertainty.

Spatial representation of terminals and radio environment

ZHAO Yuxin
Host:Ericsson
Supervisor:GUNNARSSON Fredrik
Description:New algorithms for combining information from multiple device trajectories with uncertainties in both positioning and associated radio measurements using and extending Bayesian smoothing ideas. The research task is to propose and demonstrate new information combination approaches based on a Bayesian fusion perspective.

Metadata interoperability for fusion platform

KUZIN Danil
Host:Rinicom Ltd.
Supervisor:MARKARIAN Garik
Description:To develop, analyse and optimise metadata interoperability for the developed fusion platform. The research task is to propose and demonstrate the feasibility of the proposed metadata interoperability algorithms both by simulations and practical trials.

Tracking multi-target large scale systems

DE FREITAS Allan
Host:University of Sheffield
Supervisor:MIHAYLOVA Lyudmila
Description:This project focusses on tracking a large number of targets. The objectives are to develop nonparametric Bayesian methods and combine them with sequential Bayesian methods and resolve the problems that generic sequential Monte Carlo methods face in high dimensions. New methods will be developed that are able to deal with the high state dimensions, e.g., compressed sensing, and combined with Bayesian approaches. Transformation from the high dimensional state space to a lower dimensional state can cope with the high dimensionality. These approaches will be applied to e.g., vehicular transportation systems, tracking of large groups and crowds, security and surveillance systems.

Identification of complex nonlinear systems

KASEBZADEH Parinaz
Host:Linköping university
Supervisor:
Description:The development of new algorithms for automatic calibration of unknown parameters in complex state space models, such as biomechanical systems for modelling body motion by exploiting recent developments in machine learning and convex optimization.

Efficient tracking of closely-spaced and interacting objects

MORENO LEON Carlos
Host:Thales Nederland B.V.
Supervisor:DRIESSEN Hans
Description:Tracking and labelling of objects that are closely spaced and/or interacting. Exploration and possible combination of methods like Particle Markov Chain Monte Carlo, Random Finite Set formulations and approximations for group tracking and object labelling.

Efficient state and parameter estimation in very large systems

KAMTHE Sanket
Host:University of Twente
Supervisor:MANDAL Pranab
Description:To develop techniques/methods for efficient estimation of state and parameters in complex systems. The system structures will be studied and exploited to develop efficient techniques and better algorithms for state and parameter estimation in high-dimensional systems. Proposal densities that will help with high dimensional situation will be studied and characterized for specific system structures. The idea of ensemble particle filter will also be explored to handle large systems.

Smart tracking for wide area surveillance

IGLESIAS GARCIA Fernando J.
Host:Thales Nederland B.V.
Supervisor:BOCQUEL Mélanie
Description:Development of effective methods for information extraction out of huge amounts of data gathered by modern wide area surveillance sensor systems. Efficient algorithmic solutions, e.g., based on Particle Markov Chain Monte Carlo, particle flow and sparseness will be investigated for enabling this extraction.

Multi-sensor maritime traffic monitoring and surveillance

AHMED KHAN Muhammad Altamash
Host:FKIE
Supervisor:ULMKE Martin
Description:The goal is to achieve a mid-to-long term traffic prediction and to generate a probability density map of vessel existence for maritime traffic surveillance. Combination of macroscopic/mesoscopic traffic flow models, e.g., continuum models from fluid dynamics, with Bayesian estimation theory, e.g., so-called intensity filters and context information (knowledge bases).

Simultaneous segmentation and tracking in 3D point cloud data

TUNCER Mehmet Ali Cagri
Host:FKIE
Supervisor:SCHULZ Dirk
Description:Processing of large amounts of sensor data (3D laser range sensing) for obstacle avoidance and motion control. Tasks and methodology: Development of joint classification and tracking approaches; efficient segmentation and aggregation of sensor data into the relevant objects of traffic situations.

Algorithms for practical night vision surveillance systems with large data

KOLEV Denis
Host:Rinicom Ltd.
Supervisor:MARKARIAN Garik
Description:To develop, analyse and optimise new video analytics algorithms for low resolution video streams generated during the night time surveillance with large data. The research task is to propose and demonstrate the feasibility of the proposed new algorithms. The accuracy and quality of the developed algorithms will be determined both by simulations and practical trials.

Knowledge extraction from data

ISUPOVA Olga
Host:University of Sheffield
Supervisor:MIHAYLOVA Lyudmila
Description:The objectives are simultaneous estimation and inference from heterogeneous sources. New algorithms for information extraction from complex, heterogeneous data will be developed using and extending ideas based on sparsity and compressed sensing. The research task is to propose and demonstrate new real-time data and information processing techniques using Bayesian approaches, combined with sparsity and compressed sensing.

ER Projects

Temporal representation of terminals and radio environment

YIN Feng
Host:Ericsson
Supervisor:GUNNARSSON Fredrik
Description:New algorithms for determining trend and flow properties of combined device trajectory data from a large number of devices to assess radio link performance variability over time. The research task is to propose and demonstrate new trend and flow estimation mechanisms for radio link performance including both short term and long term statistics extending the Bayesian fusion perspective to encompass also the time dimension.

Modelling and estimating complex radar objects

PODT Martin
Host:Thales Nederland B.V.
Supervisor:PODT Martin
Description:Tracking and classification of objects through the use of more complex models that will require extra state variables and parameters. Development of those complex object and observation models as well as methods for dealing efficiently with the increase in this particular form of model complexity.

Inertial body tracking

CHOW Jacky
Host:Xsens Technologies BV
Supervisor:HOL Jeroen
Description:Apply and adapt recent developments in numerical optimization and SLAM to inertial body tracking and parameter estimation. Study state of the art numerical optimization methods as well as human motion tracking; develop a robust optimization based on estimation method for large scale systems, with a focus on ensuring that inevitable modeling errors affect the state vector only locally, thus ensuring a manageable solution and validate the performance of the new algorithms on experimental data