Postgraduate Projects in Electronic and Computer Systems Engineering

PhD Projects


Groups of power consumers can be fit into demand profiles based on their behaviour. Traditionally these profiles have been created by identifying a consumer group and simply averaging consumption but demand profiles are changing due to changes in lifestyle, adoption of rooftop solar and electric vehicles. A better approach would be to identify a set of characteristic demand profiles from the data without a-priori assumptions. Given a multi gigabyte time series data set of 30 minute kWh power demands for each customer find the N characteristic demand profiles that best fit the customers. This project will develop a system that can accurately predict a N minute aggregate demand for a subset of customers that are identified using demographics and other meta data. This project is industry sponsored and supervised by Ramesh Rayudu, Paul Teal and Marcus Frean.

Evolutionary Robotics

Robots are increasingly used in everyday tasks from autonomous navigation to intelligent search and rescue. It is often not practical to pre-program the needed behaviours. It is hypothesised that evolutionary processes have led to intelligent agents. Such 'trial and error' learning has been shown possible in real robot platforms, but training these Evolutionary Computation techniques takes too long. Our group has used the EC technique of Learning Classifier Systems to share learnt information, which greatly enhances adaptation to new tasks. Improvements are sought in the areas of collective (hive) intelligence to real-world mobile robot platforms. Supervisors Will Browne and Dale Carnegie.

Affective Robotics

Humans live in a dynamic, varied and often unusual world that confuses robots that are simply reactive or preprogrammed. Affective robotics studies analogues of human emotion to enable robots to generalise and extrapolate to unknown domains. Past work in the group has shown benefits to path planning, localisation and mapping in complex tasks, but these have been in static domains. The next project is to explore dynamic domains to increase the range of likely emotions, such as anger at failing to fulfil a task. Supervisors Will Browne and Dale Carnegie.


Signal Processing

Microphones for Acoustic Holography

Holographic audio systems create an entire sound field: a much more convincing audio experience than most systems today. Recording such a sound field requires an array of many microphones. Inexpensive microphones can be used for such an array but a difficulty arises at low frequencies where phase differences between microphones account for a large proportion of the total phase variation across the array. Such variations can be corrected by a calibration procedure based on spherical harmonic decomposition. This project will develop such a microphone and an effective calibration procedure. This project is supervised by Paul Teal and Bastiaan Kleijn.

Coding Spatial Audio

Virtual reality is slowly becoming main-stream. Where-as until now the low-dimensional audio signal required a relatively low coding rate compared to video signals, the situation is very different in immersive environments. The audio signal must be known in the entire space where the person can move around in and current representations, such as higher-order ambisonics and wavefield synthesis, require very many audio signals to make that possible. The required signals display a high degree of correlation and this must be exploited. The aim of this project is i) to analyze and possibly improve the signal representations used for immersive environments to facilitate efficient coding and ii) to develop efficient coding methods for the selected representation. This project is supervised by Bastiaan Kleijn and Paul Teal


Surfaces for Enhanced Heat Transfer

Porous or structured surfaces can enhance phase change heat transfer processes and are thus becoming increasingly important in the thermal management of high power electronics. However, porous structures are highly complex and it is difficult to link different structural features to a heat transfer mechanism. This project will use microfabrication techniques to fabricate surface features with different morphologies over a range of length scales and to evaluate their efficiency and role in phase change heat transfer processes. This will aid understanding of these complex processes and help in designing more efficient electronics cooling systems. Supervised by Gideon Gouws and Ciaran Moore.

Sub-Wavelength Optical Microscopy

Optical microscopy is one of the most versatile and robust tools we have for studying the natural world around us. Unfortunately, its resolution is limited to about the wavelength of visible light, which prevents nanoscale electronic devices and smaller biological organisms such as viruses from being imaged effectively. Plasmonic devices made from metallic thin films and nanostructures present an exciting opportunity to beat this diffraction limit, opening up an optical window on a world that was previously invisible. This project will explore the design, fabrication, and integration of plasmonic devices into conventional optical microscopes, as well as potential applications for this new technology. Supervised by Ciaran Moore and Gideon Gouws.

Masters Projects

Image Processing

Characteristics of CO2 bubbles

The use of heated and humidified CO2 as an insufflation gas during heart surgery shows promise in reducing the occurrence of air embolism, thus reducing danger to the patient and saving surgical time and resources. This project aims to quantify the behaviour of CO2 bubbles during operative procedures by developing image processing techniques for tracking the location and behaviour of bubbles using transesophageal echo (TEE) images. The behaviour of CO2 bubbles in saline solution will also be studied using optical photography in order to gain insight into the fundamental behaviour and mechanisms. The project will be sponsored by Fisher and Paykel Healthcare. Supervised by Chris Hollitt and Gideon Gouws.