Systems and Control Research Group

Undergraduate Laboratories

The undergraduate laboratories service a range of undergraduate subjects for students from second year to final year. These include courses covering introductory systems and control through to more advanced studies in system identification and control, as well as courses addressing areas of real-time instrumentation and control engineering. To facilitate the undergraduate teaching program, the laboratories are equiped with desktop PC's as well as a variety of benchtop processes for practical testing of theory.

Each laboratory is equiped with PC's which are configured as dual boot machines, with Windows 2000/NT on one partition, and Linux on the other. For real-time data acquisition, processing, and control, the Real-Time Linux kernel is also installed and can take over from Linux for time critical processing. The PC's are loaded with Matlab for Windows, with Linux Matlab also available. In one laboratory, data acquisition can also be achieved in Windows
via LabView. Most data acquisition and control software for Linux, including drivers, is written in-house.

A variety of benchtop processes are used for the demonstration and practical testing of both continuous and discrete-time theory. These include a number of DC motor sets which are used primarily at the introductory level, dual water tanks, used for more advanced studies in system identification and digital control, and embedded 'Handyboard' microcontrollers, used for teaching real-time embedded system design.

The laboratories are also used by undergraduates for final year thesis projects. Processes within the laboratories which are utilised for thesis projects include: the inverted pendulum; additional balancing processes such as the ball and beam, and ball and plate; miniature overhead crane apparatus; dual water tanks for multivariable control; Mitsubishi M16C microcontrollers for embedded system design. A more recent addition is a number of different types of cameras of various sizes and capabilities to allow vision and control studies for example in conjunction with the inverted pendulum, crane and ball and plate systems. A further addition is a small hovercraft which is being developed for vision oriented autonomous navigation studies.

Chemical Process Membrane Laboratory

Academic Staff: Dr David Clements
Research Students: Edward Royston

The Process Control Laboratory is a joint facility of the School of Electrical Engineering and Telecommunications and School of Chemical Engineering Industrial Chemistry. It consists of a commercial distributed control system (MOD 300) and a pilot membrane plant, with other pilot plants currently being added. The membrane plant is designed to extract the protein from whey resulting from cheese production. The initial work is mainly focussed on the modelling and identification of the plant dynamics as a precursor to the investigation of control schemes to maximise throughput subject to fouling constraints.

Power Plant Simulation Laboratory

Academic Staff: Professor Neville Rees
Research Students: Chris Lu

The research group has for many years carried out research and application work in the field of power plant, modelling, simulation and control. The work is now centred around the Power Plant Simulation Laboratory.

The laboratory has the following facilities:

Computer Systems: The computer hardware features include: 4 Pentium III - 500MHz 128Mb DRAM 10 GHD with Zip driver and CD Writer, Printers and Scanner. The software systems are Windows and Linux. Software applications includes Matlab/Simulink and Scilab.

Power Plant Application Softwares: The centre place of this software is a 30th order nonlinear model of a 660 MW unit. The model includes the existing controls. A range of interpretive models have been developed and implemented in Matlab/Simulink. These include drum water level, superheater, deaerator, vertical spindle mill and many others. The models have been established in an overall simulation interface package for easy assembly of power plant simulation models.

Bailey Infi 90 DCS System: This system has been donated by Pacific Power. It is a fully operational small DCS system that can be linked in real time to the power plant software for studies in computer control. It can also be used to control other real time processes balancing.

The laboratory can be used by staff, 4th year thesis students and postgraduate students working in the power plant field. People doing other modelling and simulation work are also welcome to use the facilities. Recent projects include: intelligent control of coal mills, deaerator control, once through boiler modelling and simulation, superheater temperature control and feedwater heater modelling. The overall design of the laboratory and maintenance of its technical facilities is carried out by Chris Lu, ext. 5307.

Neuroengineering Laboratory

Academic Staff: Associate Professor Peter Neilson
Research Staff : Dr Megan Neilson
Research Students:
Asim Ghous, Jianming Jiang, Yalchin Oytam,
Yueshi Shen

What is Neuroengineering?
Neuroengineering is a new field that stands at the interface of neuroscience and engineering.

Neuroengineers are interested in the development of brain-like computational systems that emulate significant aspects of real-world adaptive learning and control. To do this they combine the study of engineering techniques with the study of biological neural systems.

Kumpati Narendra, Professor of Electrical Engineering and Director of the Neuroengineering and Neuroscience Centre at Yale University, describes the philosophy of Neuroengineering as follows:

“Biological systems contain millions of examples of both simple and complex nonlinear multivariable systems which are controlled adaptively. In fact, the tremendous efficiency of control strategies to be found in the biological world has inspired engineers for centuries and has led in recent decades to the appropriation of numerous terms, such as adaptation, learning and self-organization from psychology and life sciences, for use in engineering. The analysis and synthesis of such systems will require all the tools in the arsenal of the control engineer, including adaptive and learning techniques, stochastic control, optimal control, automata theory and game theory. It is anticipated that the increasing interaction between adaptive control theorists and neurophysiologists will lead to better and novel design of neural adaptive controllers for nonlinear, multi-input, multi-output systems.”

K.S. Narendra (1995), Adaptive control: neural network applications. In M.A. Arbib (Ed.), The Handbook of Brain Theory and Neural Networks. The MIT Press.

Neuroengineering at UNSW

In 1998 UNSW established the Neuroengineering Laboratory in the Systems and Control Research Group, School of Electrical Engineering & Telecommunications.

The laboratory specialises in the study of adaptive optimal control of multivariable nonlinear dynamic systems, using the human motor control system as an experimental paradigm. It fosters the establishment of multidisciplinary links and a shared knowledge base across the UNSW community for theoretical and applied research on multivariable nonlinear dynamic systems in a variety of fields. As well as involving engineering disciplines such as robotics, the work of the laboratory includes aspects of computational neuroscience, systems physiology, human information processing, human performance and sports science. There is special interest in the study of neurological disorders of movement; e.g., cerebral palsy, stroke, Parkinson’s disease and stuttering.

As well as offering postgraduate training, the Neuroengineering Laboratory makes project work available to undergraduate students in their final year. These students become directly involved in the experimental and/or simulation work being undertaken in the lab, and consequently gain hands-on experience of real-life research and development in a multidisciplinary setting.

What do we do in the Neuroengineering Lab?

Experimental workstations feature dual computers equipped with real-time dynamic system simulation and Biopac data-logging hardware and software. This provides great flexibility and allows a range of electrophysiological variables (e.g, muscle activity EMGs, heart activity ECGs, brain activity EEGs, eye movement EOGs) as well as joint-angles and/or postural balance forces to be monitored on-line.

Most experiments in the lab require human subjects to perform some sort of tracking task. Using the above equipment, we can create an immense variety of tasks, many of them unusual and challenging. A ‘generalised tracking software package known as Hypertrack has been developed and this is integrated with MATLAB analysis and simulation software within each workstation. The tracking program effectively turns the workstation into a simulated multivariable dynamic system which can be controlled in real-time by the subject at the same time as physiological variables are monitored and stored. The program allows any measured variable to be used to provide real-time control of the simulated system. The multivariable responses from the simulated system can be displayed to the subject in real-time via a variety of devices, and can involve visual, auditory, and/or kinaesthetic systems. The tracking program generates target waveforms that are presented to the subject via the same displays. The task for the subject is to make the responses track the target waveforms as accurately as possible despite disturbances and time-variations programmed into the system. By employing time series analysis of the tracking variables, the ability of the subject to control various physiological responses (e.g., a limb movement as measured by joint angle goniometers, or the strength of a hand grip as measured by force transducers, etc.,) can be assessed quantitatively and described in terms consistent with those employed in the engineering analysis of multivariable feedback control systems. This allows us to study the adaptive control characteristics of various sensory-motor systems and in turn to develop biologically plausible computer simulations of how the brain controls movement.

Below is a SIMULINK simulation of human tracking performance that includes adaptive learning, prediction of target movement and compensation for predicted disturbance to the response.

Biomedical Systems Laboratories

Academic Staff: Professor Branko Celler,
Research Students:

See Biomedical Systems Laboratories

Vision and Control Laboratory

Academic Staff: Professor Victor Solo,
Research Students: Vacancy

This lab is under construction. Facilities include a number of PCs and UNIX workstations, specialised cameras and a small hovercraft. Current research activities include: motion estimation from image sequences; autonomous vision and control of a hovercraft; identification of distributed systems; wireless navigation; adaptive signal processing in Telecommunications. The research is mostly supported by ARC Discovery Grants.