Projects

Leader: Brian Anderson

The Cooperative Multi-Agent Formations and Networks project, known by its nickname SWARM-II, focuses on research under four main topics

1. Network functional health, robustness and capabilities;

2. Development of discrete probability density maps (DPDM);
3. Time difference of arrival de-interleaving for localisation;
4. Development of general tools for cooperative target localisation and sensor self-localisation, including when GPS is lost, and determination of localisation performance.

Each topic bears on various fundamental research problems driven by certain real life scenarios of DSTO interest as well as more specific DSTO research tasks, as represented by the second and third topics. There is significant collaboration with DSTO scientists; this has recently expanded beyond Electronic Warfare and Radar Division (EWRD) to include Intelligence, Surveillance and Reconnaissance Division (ISRD).

While the primary user of the research is Australian Defence, the US AIrforce has been providing and is continuing to provide funding through AOARD, to focus on robustness of sensor network localization and formation control, especially in the face of agent or link losses. The work has other potential users, e.g. those with interest in monitoring bushfires or surveillance tasks and companies such as CEA Technologies, with whom some collaborative activity has taken place.

The longer-term plan is to retitle this project to DICE (Distributed Control and Estimation in Networked Environments), and evolve the studies to include more work on information fusion including scheduling, and on modelling and design of highly mobile networks. This will lead to a close collaboration between this project and the Projects below.

Large Scale Distributed Estimation and Tracking (LSDET)

Leader: Subhash Challa

This project addresses some of the fundamental problems associated with distributed object tracking and feature estimation in large scale, distributed and networked sensor systems. The problem arises in closed world scenario, where sensors observe objects entering and leaving the surveillance from all access points and open world scenario, where the sensor coverage is partial. The objects have a multitude of features that are projected into the sensor measurement space and lead to either fully or partially observable systems. Moreover, no single sensor can cover the complete surveillance region – hence there are number of issues relating to object handovers and estimated states over band limited communication channels. The focus on this project will to address these problems with special focus on:

1. Networked tracking and fusion over bandwidth limited channels
2. Distributed tracking and fusion algorithms for scalability over network size & complexity
3. Sensor-Object Registration and Self-Calibration across large scale distributed networked sensors
4. Data Association and tracking across multiple over lapping and non-overlapping sensors.

Solutions to these problems have great significance to practical problems in Defence, Surveillance and Security industries.

Decentralised Control of Very Large Scale Systems via Optimisation (DCO)

Leader: Rob Evans

This project addresses the challenging problem of designing and implementing optimal control strategies for very large distributed dynamical systems supported by networked communication infrastructure. Such systems are characterised by vast numbers of networked sensors and actuators each measuring and controlling small parts of the total large dynamical distributed system. The use of emerging powerful optimisation tools will be explored as an approach for implementing decentralised moving horizon control strategies.  This whole area is in its infancy and so far only initial approaches have been suggested for designing and implementing optimisation based control methods for the implementation of decentralised control.  

Developing the foundations and practical techniques for filtering information from the large volumes of data produced by emerging high-throughput biomedical technologies, in particular high throughput sequencing technologies, often referred to as Next Generation Sequencing (NGS), but also a range of other, more cost effective technologies.