Electrical and Computer Engineering

Exploring augmented and virtual reality

Supervisor

Dr Nasser Giacaman
Faculty of Engineering
Project code: ENG043

Augmented reality (AR) and virtual reality (VR) technologies are gaining popularity in mainstream applications. This project will involve designing and building either an AR or VR mobile application in an educational domain. The particular domain selected will be determined closer to the time, but may include helping students learn mathematics or programming concepts. To be successful in this project, you should be a strong programmer confident in using Unity (C#).

Exploring web-based learning resources

Supervisor

Dr Nasser Giacaman
Faculty of Engineering
Project code: ENG044

This project will involve designing and building web-based extensions to a learning resource. The particular domain selected will be determined closer to the time, but may include helping students learn mathematics or programming concepts. To be successful in this project, you should be a strong programmer confident in using JavaScript and other web-based technologies.

Designing and testing of IPT pads for in-motion charging of Evs

Supervisor

Dr Seho Kim, Prof. Grant Covic
Faculty of Engineering
Project code: ENG045

In-motion charging of electric vehicles (EV) using inductive power transfer (IPT) has become a mainstream research topic recently. The goal of this project is to design and develop a scaled down in-motion IPT pad, which will be tested for thermal and mechanical stress.

Applicant preferably has some experience/knowledge in electrical circuits and magnetics.

Simulation of IPT pads for in-motion charging of Evs

Supervisor

Dr Seho Kim, Prof. Grant Covic
Faculty of Engineering
Project code: ENG046

We are looking for a motivated summer research student willing to take on a project related to inductive power transfer (IPT) systems during the summer.

The student would be tasked with:
- Design of magnetic components using a CAD software, such as Solidworks
- Simulation of the magnetic components in ANSYS Maxwell
- Integrating the simulation models with other simulation packages for thermal and mechanical analysis
- Validation of simulation model to a scaled prototype

Skills required: Knowledge in magnetic simulation packages, such as ANSYS or JMAG, and Solidworks would be preferable, but not necessary.

A model in-motion charging system

Supervisor

Dr Duleepa J Thrimawithana
Faculty of Engineering
Project code: ENG047

In-motion charging of electric vehicles (EV) using inductive power transfer (IPT) has become a mainstream research topic in recent times. The goal of this project is to design and develop a scaled down in-motion charging system that may be used in ECSE courses for teaching as well as demonstrations.

During this project, the student is expected to design and develop the power electronic converters and magnetics required for the power transmitters. The student should also develop a lightweight track with guide rails so that a remote-controlled car can be driven along the track to demonstrate in-motion charging.

The initial system specifications are:
- Supply voltage to transmitter - 40 Vdc
- Maximum transmitter power - 50 W
- Track length - 10 m

Skills required: Students should have experience/knowledge in Altium Designer, electromagnetics, power electronics, simulation and CAD.

Bi-directional wireless power transfer demonstration system with USB-C capability

Supervisor

Dr Duleepa J Thrimawithana
Faculty of Engineering
Project code: ENG048

The aim of this project is to develop a system to demonstrate the ability to transfer power wirelessly between two devices. Both devices will have USB-C ports, that enable bi-directional power flow between them through the wireless power transfer system. During the first stage of the project, the student is expected to build a 50 W bi-directional inductive power transfer system. The second stage of the project involves adding UCB-C capability so that the bi-directional inductive power transfer system can be interfaced with any USB-C device. The functionality of the system will be demonstrated using two power banks.
The initial system specifications are,
- Power rating - 50 W
- Rated voltage – 5, 12 or 20 V
- Power transmission distance – 30 mm

Skills required: Students should have experience/knowledge in Altium Designer, electromagnetics, power electronics, simulation and embedded programming.

Evolving neural networks for reinforcement learning environments

Supervisor

Dr Henry Williams
Faculty of Engineering
Project code: ENG049

This project forms an extension to the on-going SFTI robotics spearhead with the CARES group at the University of Auckland. The core project is seeking to develop a means of enabling robots to learn to interact with their environment without a human in the loop.

The selected student will seek to develop an evolutionary computation approach to generating neural networks that are capable of learning in the domain of reinforcement learning. This approach will be evaluated against the OpenAI Atari 2600 gym to determine its feasibility, before extending and applying it to a robotics application.

This project requires strong skills in programming, in particular Python. Experience in machine learning/tensorflow will help but is not required.

Utilising an agent's learned understanding of the world to learn a new task

Supervisor

Dr Henry Williams
Faculty of Engineering
Project code: ENG050

This project forms an extension to the on-going SFTI robotics spearhead with the CARES group at the University of Auckland. The core project is seeking to develop a means of enabling robots to learn to interact with their environment without a human in the loop.

Ideally, a robot would be able to understand the consequence of its actions with the environment. Using this understanding the robot would be able to plan or learn to complete tasks in its own “mind” before having to interact with the environment itself. Potentially saving training time, and reducing the risk of harm from learning in the real-world.

The selected student will investigate a means of utilising the current state of model-based reinforcement learning such that a robot can learn a new task from the learned model/learned understanding of its interaction with the environment.

This project requires strong skills in programming, in particular Python. Experience in machine learning/tensorlfow will help but is not required.

Minimising energy consumption in multi-core hard real-time systems

Supervisor

Prof. Zoran Salcic
Faculty of Engineering
Project code: ENG051

Increasing reliance on computer power leads toward high energy consumption especially in embedded real-time systems that operate non-stop. These systems rely on multiple (sometimes heterogeneous) cores implemented in the same chip. We have worked on aspects of hard real-time systems designed by using SystemJ language that extends Java to synchronous and GALS (Globally Asynchronous Locally Synchronous) model of computation and their implementation as multiprocessor system on chip. Energy consumption was not considered as the functional and timing correctness were the primary goals. In this project we extend the considerations to energy consumption, thus leading towards hard real-time systems with low energy consumption, which preserve their hard real-time behaviour.

Skills: Good understanding of computer architecture and multi-core systems, programming in high-level languages (C/C++ and Java), familiarity with embedded and real-time system concepts, basic knowledge of FPGAs and exposure to Intel/Altera Cyclone family preferable.

Taming the energy consumption in linear arrays of wireless sensor nodes

Supervisor

Prof. Zoran Salcic
Faculty of Engineering
Project code: ENG052

Sensing technologies will be used in an increasing number of applications helping monitor physical processes and operations and making detection of faults and maintenance easier while better utilising the resources involved. Some of these physical systems (such as roads, pipelines, bridges etc.) are stretched over bigger distances and naturally lend to the use of wireless sensor networks with linearly allocated nodes (chain-type topology).

In this project we look to the specifics of such sensor networks under assumptions that all nodes cannot be provided permanent power supply and have limited transmission range, but the information collected by the nodes need to be available at some central point for processing and interpretation.

Skills: Ability to model communication protocols in OMNET++, programming embedded platforms (wireless sensor nodes) in C.

Data processing for error detection

Supervisor

Dr Kevin Wang, Dr Andreas Kempa-Liehr
Faculty of Engineering
Project code: ENG053

In this project, the topic of data uncertainty will be explored by investigating various issues that occurred during IoT (Internet of Things) data collection. Multiple public and/or private dataset will be studied (extracted, cleaned, and/or labelled), and feed through different types of machine learning algorithms to evaluate the impact of those issues and the effectiveness of the used algorithms in detecting those issues.

Students interested in data processing and analyses, with good programming and reasonable machine (or statistical) learning abilities are welcome.