Centre for Intelligent Systems Research

CISR Research Seminar Series - 2009

  CISR presentation
blue star Professional development
yellow star Keynote lecture
purple star External presentation
red star No presentation

These research seminars were held in the Geelong Haptics Lab (KC1.017)

Date Time Presenter Presentation/topic
Thursday October 15 2pm Ali Ghanbari purple star  Neuro-Fuzzy Microrobotic System Identification for Haptic Intracellular Injection 


Using a haptic interface to control a micromanipulator, which is critical in intracellular injection, has many beneficial implications. In particular, the haptic device should be able to control the microrobot in Ám resolution, requiring an accurate model of the system. As the system has an unknown internal structure with a nonlinear behaviour, a neuro-fuzzy dynamic model has been developed. Adaptive Neuro-Fuzzy Inference System (ANFIS) is employed as the system identification approach to model the non-linear dynamics. The developed ANFIS model is able to predict the microrobotic system response very precise with root mean square error of 1.0224 Ám while the microrobot manoeuvres in cm range (104 times).

Thursday October 1 3pm Mats Isaksson Improving the Kinematic Performance of the SCARA-Tau PKM 


One well acknowledged drawback of traditional parallel kinematic machines (PKMs) is that the ratio of accessible workspace to robot footprint is small for these structures. This is most likely a contributing reason why relatively few PKMs are used in industry today. The SCARA-Tau structure is a parallel robot concept designed with the explicit goal of overcoming this limitation and developing a PKM with a workspace similar to that of a serial type robot of the same size. This paper shows for the first time how a proposed variant of the SCARA-Tau PKM can improve the usability of this robot concept further by significantly reducing the dependence between tool platform position and orientation of the original concept. The inverse kinematics of the proposed variant is derived and a comparison is made between this structure and the original SCARA-Tau concept, both with respect to platform orientation changes and workspace.

Thursday September 10 3pm Wael Abdelrahman Interactive simulation of Deformable Objects 


Haptic rendering of deformable models requires physically accurate simulation to reflect the applied external forces. This is a nonlinear simulation due to the object material, topology, and the external applied force. Accurate techniques such as finite element methods (FEM), to calculate such simulations were introduced but still they cannot be applied in interactive haptic applications. On the other hand, training methods and problem reduction approaches were introduced to generate quick solutions. Those methods rely on acquiring information about the underlying object to guide a problem reduction technique or to be fed to a learning process. The results can then be used to produce interactive deformable models simulation. We propose novel techniques to automatically generate the training sets or the deformation basis. The goal is to deal with the nonlinearities of the problem and produce adaptive algorithms that are flexible according to the required number of training sets or the available resources.

Thursday August 27 3pm Matthew Watson Creating a Sparse 3D Feature Model using Short Baseline Stereo & Multiple 2.5D-View Registration  


Having some level of a-priori information about an object is very useful when detecting and estimating its pose in real-time systems, i.e. augmented reality. A-priori information generated from training data must accurately represent distinctive and repeatable object features, and be robust to scaling, illumination and perspective distortion. This talk will outline current work in developing a algorithm to generate a distinctive object representation offline, using short baseline stereo fundamentals to triangulate highly descriptive object features in multiple pairs of stereo images. When a set of sparse 2.5D perspective views are built, the multiple pairs are then fused into a single sparse 3D model using a common 3D shape registration technique.

Thursday August 13 3pm Khashayar Khoshmanesh Development of Magnetic Microrotors for Active Micromixing 


This work presents the development of a 3-D active micromixer. The mixer utilises two magnetic rotors to agitate the entering flows. Each rotor has three blades connected through a hub. The width of the blades increases linearly to enhance their impact on the surrounding flow. The performance of the mixer is studied within the rotation speeds of 150-275 rpm. Results indicate that the mixing efficiency increases linearly from 88.91% to 93.58% within the range of 150-250 rpm. Further increasing of rotor speed does not improve the mixing due to the resistance of corner flows.

Thursday July 30 3pm Carl Jiang Data modelling based on SOM and Inverse Problem theory 


Discovering knowledge in measured data is very important for us to determine data input model for modelling and predicating passengers and handbags at a specific section of airport. The unique technique - SOM (Self-Organising Map) will be introduced to classify and visualise raw data collected from different sections of airport, such performance is useful for us to gain prior information from raw data for selecting candidate model. Based on obtained prior information and experience of selecting suitable parametrical model for predication, the challenging technique - inverse problem theory will afterward be introduced for estimating values of parameters of candidate model and eventually confirming data input model with higher accuracy. Meanwhile, some detailed examples of how to solve cases of engineering or science using SOM and inverse problem theory will be represented as an approach to further research in this field.

Thursday July 9 3pm Abbas Khosravi Developing Hybrid Modeling and Simulation Methodologies for Complex Systems 


This presentation will outline findings of my research within the last year. First, common techniques for developing abstract models of complex systems will be briefly reviewed. Especial attention will be paid to nonlinear non-parametric techniques, including neural networks, for abstracting complex systems. Prediction intervals, rather than point prediction, will be introduced as a promising tool for dealing with uncertainty of complex systems. A variety of techniques for constructing prediction intervals for neural networks will be reviewed. Some preliminary results will be demonstrated for constructing prediction intervals for outputs of complex systems. Finally, potential domains for improving performance of these techniques (e.g., reducing length of prediction intervals without compromising their coverage probability) will be discussed.

Thursday June 25 3pm Hamid Abdi Cooperative and Dynamic Multi-Robot Multi-Target Mission and Path Planning Using Potential Field Concept  


Mission and path planning for multiple robots in dynamic environments are required when multiple robot or UVs (UAVs, GUVs, UUVs) are used and a group of targets needs to be done. The problem is critical for most multi cooperative robots in dynamic environment as an instance a group of rescue robots or UAVs for rescue needs mission and path planning. This problem is a complex decision making problems which includes different uncertain environmental, obstacles and robotic issues. In this study different cases of the problem have been solved and will be introduced and the most complex form will be discussed. In this form more over than potential field concept the priority for robots, targets are considered by introducing suitable weighting coefficients. Uncertainty in movement and threatening concept has been used for obstacles. Finally the solution will be presented by a simulation in dynamic way, means for every run because of properties of dynamic environments and inherent uncertainties, the provided result will be different. To show how this method can be used under various optimality conditions, we developed region priority concept for robot movements which will be indicated.


Deakin University acknowledges the traditional land owners of present campus sites.

27th March 2012