Future projects

With the nature of our work here at the Institute for Intelligent Systems Research and Innovation (IISRI), we never stop looking to the future and how our work can make that future smarter and easier.

What's ahead?

We're currently working on a whole suite of projects across our five main streams of research – from a haptically enabled interactive living room to mitigating consumers everyday energy use. Browse the topics below and get inspired by a research-driven future.


Haptically enabled interactive living room

This research aims to investigate how to expand low-cost motion tracking gaming consoles to improve user interaction in the living room. An occlusion-free multi-hyper skeleton motion tracking system will be designed and implemented. The anticipated outcomes of this research will facilitate living room rehabilitation, social networking, micro-targeting advertising and autistic recovery.

Vibrating tactile tracks for facilitating dancing experience to visually impaired audience

This research aims to provide a visually impaired audience with the ability to perceive and understand dance movements. The research takes place in two directions. First, introducing tactile track to composers and choreographers. This track will require a vibrating hardware fitted in the venue seats. The second direction is facilitating dance to visually impaired actors. This will require a wearable hardware to be fitted on the visually impaired actors.

Developing modelling methodology to determine critical process parameters and material properties

This project aims to address limitations of the design of experiments methods. We'll do this by developing a modelling methodology and tools for the process optimisation based on the theory of non-linear dynamical systems, including excitability and attractors. In particular, our aim is to determine the desirable process attractor in regards to process parameters. As well as the material properties that would allow researchers to quickly and accurately determine critical process parameters and optimal process settings in regards to product quality and/or productivity.

Force distribution analysis

This project involves the analysis of haptics devices manufactured by different vendors. This analysis and the relevant framework developed will provide a benchmarking tool to compare different devices in terms of quantitative and qualitative performance. This research will further generate a performance matrix highlighting the suitability/feasibility of existing off-the-shelf devices for any particular application. The factors involved in the performance comparison are peak static and peak continuous force reflection, work envelope, force and position control, accuracy and resolution etc. In addition, this research will provide grounds for the validation of vendors' claims about the device.

Haptics modelling of flexible materials

This project involves the development of a framework and methodology that will allow interaction with, and manipulation of, flexible materials in real time. The research will reflect the properties of the materials, such as tension, torsion, bending, surface friction and forces at any certain point. Currently, the focus of this particular research is the modelling of 1D objects such as cables, hoses and harnesses. This focus will be extended to 2D analysis, providing grounds for the modelling of visco-elastic materials as an end goal.

Haptics-enabled nano-manipulation

This project involves the development of a methodology and framework to interface haptics technology with an atomic force microscope. The end product of the research will open new horizons for nano-scientists and will provide a direct and intuitive way of nano-manipulation of materials at an atomic level. The haptics system will reflect all the interactive forces exerted on the cantilever tip of the microscope in real time during manipulation.

Algebraic computing (programming algebra)

Scientific computing is the field that studies scientific phenomena, models them into computer and mathematical systems, and adapts these models to design new families of algorithms, data structures and mathematical solutions. Ant colony optimisation, genetic algorithms, neural networks and simulated annealing are all typical results of scientific computing theories.

However, the objective of all these solutions is mainly optimisation, tuning and searching for the best approximate solution. They simulate the problem on the physical level rather than the semantic and understanding level. This is why adapting these methods to handle and learn highly semantic problems is challenging.

This study uses soft computing algorithms/structures and upgrades them into algebraic ones to maintain high semantics interpretation. It develops an algebraic framework that binds the concepts of genetic programming (not algorithms), agent programming and concept-oriented programming to span a space of programs. This research provides a new area of understanding algorithms. It allows analytical study and design of algorithms and data structures.

A program's space, spanned by algebraic structures defined on a set of independent concepts, provides a framework for designing self-repairing programs (interpolation) and generating new programs (extrapolation). Program hyper-surfaces can be derived and searched for local minimums where the optimum algorithm/data structure lies.

Augmented Reality Mark-up Language (ARML)

This research extends the known techniques of augmented reality into a generic framework for developing augmented reality applications, scenes and media production. ARML is a tag language through which coloured barcodes/tags can be translated into 3D objects, animations and movies on top of a newspaper, a book or in a classroom.

The research is subdivided into three phases. First, a generic framework for augmented reality is to be defined and implemented. This phase includes identifying a basic instruction set. Then, a coloured barcode is to be developed for each instruction, system hardware is to be put together, and software API and rendering engines are to be developed on mobile phones. Finally, OCR features can be added to allow the user to search through the text of a book or newspaper.

Implicit 3D graphics processing for viscoelastic haptic rendering (graphics and haptics)

Efficient collision detection is one of the very challenging problems facing force feedback (haptic) rendering. Researchers managed to achieve efficient runtime rendering using optimised data structures, caching, force shading and hybrid surface representations. However, these solutions lack the ability to adapt with deformable surfaces due to the time overhead to be spent on restructuring the geometry representation in the memory.

Multi-point collisions, surface deformations and multipoint haptics (1000 fps each) add more complexity to the problem and put conventional data structures techniques out of business. Implicit surfaces had a potential in representing, deforming and recovering 3D meshes. The main benefit of using implicit surfaces is developing one equation that represents the whole surface, which means that collision points are where these equations intersect. Solving the equations of two surfaces does the job. Furthermore, incorporating conventional structuring and rendering techniques minimises the complexity of the surface equation.

Continuous image algebra: from finite sets to posets and cyclic structures (image processing)

This research studies heterogeneous image fusion from an algebraic perspective. It develops a heterogeneous algebra for fusing images, colour maps, interpretation, properties and dimensionalities. The framework also declares limits of the fusion process and a set of constraints that governs the duality between fusion operators and quality metrics. It allows a criteria upon which an automated algorithm can select the quality metric that best suits every fusion algorithm and vice versa.

Robotics and motion simulation

Force controlled body weight support system for lower limb rehabilitation

This PhD project will investigate novel design and improvements of a body weight support system (BWSS) to achieve higher performance and force control. This research will study new methodologies for the lower limb robotic rehabilitation using a system with the ability to control the weight that the person must endure. 

Several methods for BWSS are going to be studied, allowing the applicant to choose the best option and improve it, and to control the weight of the patient.

Synchronised and cooperative control system for a lower limb robotic rehabilitation

The main goal of this proposed PhD project is to design a synchronised cooperative control system of modular robots for lower limb rehabilitation tasks. The mechanism will consist of three adjustable force controlled robots that cooperatively help the patient in walking stability and leg motions. 

The research will derive a mathematical model for the lower limb robotic rehabilitation system, then the derived model will be utilised to simulate, and finally design, a control system.

A cooperative robotic system for lower limb rehabilitation

The aim of this PhD project is to design a cooperative and modular robotic system for lower limb rehabilitation. The mechanism will consist of three adjustable force controlled robots that cooperatively help the patient in walking stability and leg motions. In this project, the PhD candidate will design, simulate, evaluate and finally fabricate the complementary mechanical and electronic parts of the system to use robots for rehabilitation applications.

Robust adaptive control approach of bilateral tele-operation of tele-surgical systems

The aim of this research project is to tele-operate a semi-autonomous surgical system using a dynamic, robust, adaptive control approach. The trajectory and force profile of the surgical system is going to be commanded by the remote operator. The controlled system must have two level control systems. The higher level controller should produce the required path and force profile signals as the control commands for the system. The lower level controller transforms the higher level control signals to the motor voltage, current or frequency commands respectively.

Improving reliability and safety of robotic motion simulators using developing robust control systems

This PhD project aims to investigate robust control strategies of IISRI's motion simulator. In order to do this, appropriate fault prediction and identification features are developed for the motion simulator and high-level robust control systems will be implemented. An artificial intelligence approach will be deployed for development of fault prediction and identification. A robust control approach will be used to implement high-level motion planning and control for the motion simulator.

Fault-tolerant for medical robotics by using model predictive control methods

In this PhD project, the applicant studies fault-tolerance for medical robotics by using model predictive control methods and considers both sensor and joint failures. The aim is to enable a medical robot to perform a task. This includes maintaining the path and compliance or performing safety routines for any fault scenario to ensure the human safety during robot operation. The project will result in an improved robot control system with higher reliability and safety.

Improving householder energy consumption behaviour using data analysis and pattern recognition techniques

In this research, the applicant will use data analysis and pattern recognition methods for analysing the energy consumption profile of householders in order to identify wrong/low efficiency energy consumption behaviours. The results will then be used to develop appropriate strategies to improve the energy efficiency of householders.

fNIR spectroscopy for brain computer interface

This research aims to study functional near infra-red brain signals in order to design an intuitive brain computer interface (BCI). The anticipated outcome of this research is a modular BCI system that communicates with Google GlassesTM. Applications are limitless. The analysed signals will facilitate interaction with wearable computers, wheelchair control, and provide a safer driving experience.

Fusing ultrasound and X-ray imagery for a minimal radiation fluoroscopy

This project aims to fuse high-resolution detailed X-ray images with live ultrasound videos to produce a live fluoroscopic experience with minimal radiation doses. The proposed algorithm will enhance the freeze-frame technology by minimising the rate at which frozen X-ray frames are obtained during a fluoroscopic-guided procedure. 

This research will have great impact on minimally invasive fluoroscopic interventional procedures carried out on the abdominal part of the human body. This impact can be measured in terms of reduced radiation doses or lengthened fluoroscopic procedures at the same radiation level of current fluoroscopic systems. The anticipated outcome of this research will also open doors for carrying out safe CT-fluoroscopic procedures.

Pairwise client side image stitching via bluetooth ad hoc networks

This project aims to collect images from different mobile devices recording a particular phenomenon. The main outcome is a bigger picture publicly available for all participating mobile devices. The main motivation is to capture every little detail monitored by the crowd witnessing this phenomena. The project has a potential to be scaled up to maintain videos. Applications are limitless. They range from parties, concerts, natural phenomena, riots and revolutions. The outcome of this project can also facilitate situational awareness for autonomous mobile robots.

Micro/nano system modelling and manipulation

Analyse and understand the effect of neuroplasticity on neural information flow

This research is focused on the design and understanding of the influence of neuroplasticity on the behaviour of neuron networks and the flow of information. The research will also cover how neurons reconnect themselves when there is a network breakdown, such as after physical injury. Multi-electrode technology will be used for this research and will have significant impact on the efforts of neuro-rehabilitation after physical head injuries.

Design of novel neural interface with enhanced signal-to-noise ratio

This research involves the design and development of innovative micro-electrode arrays to interface with living neurons. Current micro-electrodes use planner structures, which are relatively easy to fabricate but lead to very low signal-to-noise ratio. This compromises the accurate detection of the neural activities. 

This research is about the optimisation of the parameters that could lead to higher signal-to-noise ratio. It involves the enhancement of geometric structures of the electrode, electrical conductivity and endurance of the electrode material, adhesion with the cells and the electrode density for better neuron-to-electrode mapping.

Unified spike detection and classification framework for multichannel recording

In vitro multi-channel recordings from neurons have been used as important evidence in neuroscientific studies to understand the fundamentals of neural network mechanisms in the brain. Accurate detection and sorting of neural activity waveforms becomes a key requirement for creating meaningful machine brain interfaces and understanding the working principles of neural networks.

In this work, we propose a unified framework for unsupervised neural spike clustering. The proposed framework exploits the features of wavelets scale-space representation and time-frequency localisation. Plus multi-scale principle component analysis to minimise the dimensionality of the raw data at different scales prior to clustering.

Accurate neural spike classification using living cells distribution around the electrode

In vitro multi-channel recordings from neurons have been used as important evidence in neuroscientific studies to understand the fundamentals of neural network mechanisms in the brain. Accurate detection and sorting of neural activity waveforms becomes a key requirement for creating meaningful machine brain interfaces and understanding the working principles of neural networks. 

This work is about the design of new classification algorithms to employ the prior knowledge of neuron cells distribution around the electrode in N sources to 1 sink information flow configuration. This research will enhance the quality of understanding of the information flow using multi-electrode architecture.

Mapping and prediction of dynamic information flow through cortical neural networks

This research involves the design and development of novel experimental set-up and algorithms to map and predict neural behaviour and information flow in response to external stimulations. It will provide insight into the activity of the brain and will shed light on how neurons pass information through the network and what makes them stop communicating.

Process modelling and analysis

Enabling ambient intelligence for manufacturing processes through distributed camera networks

This project will develop methods to optimise and schedule networks of smart and traditional cameras in a manufacturing environment, enabling knowledge capture, managing performance and identifying causes of quality degradation. This research will assist Australian manufacturers to stay competitive in the dynamic global market.

Metamodelling and optimisation of complex systems

This project will utilise event simulation-based meta-modelling capability, coupled with optimisation, to address the challenges of systems optimisation and simulation challenges for regional decision-makers. The research challenge is in accurate estimations, time series prediction and the integration of effective optimisation methods. Such problems in most real-world applications are large scale and may involve non-smooth, non-convex functions.

FEA optimisation and visualisation

The study of constitutive laws and physical behaviours of advanced materials, from organs and skin to nano surfaces, needs to consider non-smoothness and multi-scale effects. Modelling, design and simulation of these advanced materials must deal with non-convexity and large-scale deformation, which produce fundamentally challenging problems in both theoretical analysis and scientific computations. This investigation will be coupled with advanced visualisation for effective communication and analysis.

Decision support using optimisation, machine learning and constraint programming for supply chain management

Regional industries face challenging supply chain problems. From the minimisation of inventory costs to demand forecasting and adjustment, matching sales to capacity, production planning and, not least, transport planning and scheduling and environmental impact. The research challenges in optimising regional supply chains include modelling, reinforcement learning, continuous and discrete optimisation, and the integration of optimisation methods for the different sub-problems.

Construction of prediction intervals

The aim of this research program is to investigate applicability of, and exploit, advanced artificial intelligence (AI) methods for uncertainty quantification. Uncertainties associated with values predicted by AI models will be quantified and measured through the construction of prediction intervals. 

This research project focuses on developing new techniques for generation of reliable PIs using AI methods, such as neural networks and fuzzy systems. In particular, advanced type-2 fuzzy logic systems will be investigated due to their excellent capability in dealing with uncertainties.

Optimisation of prediction intervals

The aim of this research program is to investigate applicability of, and exploit, advanced artificial intelligence (AI) methods for uncertainty quantification. Uncertainties associated with values predicted by AI models will be quantified and measured through construction of prediction intervals (PIs). 

This research project considers how to improve the quality of PIs. The idea is to develop PIs that are more informative (narrower) and theoretically correct (a coverage probability above the confidence level). The PI quality improvement is challenging, as the optimisation problem is multi-objective (formulation and solution).

Application of prediction intervals

The aim of this research program is to investigate applicability of, and exploit, advanced artificial intelligence (AI) methods for uncertainty quantification. Uncertainties associated with values predicted by AI models will be quantified and measured through construction of prediction intervals (PIs). 

This research project will examine how PIs can be used in real-world decision-making processes. PIs will be used for operational planning and scheduling in a variety of fields, such as manufacturing, energy systems, transportation system and logistic networks.

Contact us

Professor Saeid Nahavandi
Email Professor Nahavandi

Deputy Director
Professor Douglas Creighton
Email A/Prof. Creighton