Mechatronics and Machine Vision in Practice, 2006

Toowoomba December 5-7

Day 1.  Tuesday 5 December

Opening address 9.45 - 10.00

Viewpoint          10.00 - 10.30

018: Click on number to see the full text.

Coordination in Mechatronic Engineering Work

James Trevelyan
School of Mechanical Engineering, The University of Western Australia

This paper shows that little has been written on the roles that people actually perform in the course of mechatronics engineering work.  The paper reports empirical results of interviews with several engineers working in mechatronics: this is part of a larger study of engineering work in several disciplines.  The paper argues that coordinating the work of other people is the most significant working role, both in mechatronic engineering and also other engineering disciplines.  This role has not been explicitly identified before.  While coordination appears to be a generic non-technical role, close examination reveals that technical knowledge is very important for effective coordination in engineering work.  The coordination role is not mentioned in engineering course accreditation criteria. The absence of explicit references to this role in previous literature suggests that the research methods used for this study could provide better guidance for engineering educators on course requirements.

Education 1        11.00 - 12.30


Design and Implementation of a four rotor helicopter and study of its stability

J. Toledo, M. Sigut, L. Acosta, J. Felipe, S. Torres and E. J. Gonzalez
Department of Systems Engineering and Control and Computer Architecture,
University of La Laguna, Spain

In this paper, a study of a four rotor helicopter prototype control is presented. Firstly, the different aspects of the prototype construction will be described, making an special emphasis in the mechanics, the design of the sensorial and actuation systems and the prototype control. Next, the formalism of the lifting and inverse lifting operators will be presented. As it will be shown, they can be applied to this multifrequency system in order to allow to use a simple lineal method to analytically determine the closed-loop system stability.

Emergent Behaviour Real-time Programming of a Six-Legged Omni-Directional Mobile Robot: Planning of Viennese Waltz behaviour

Frank Nickols
Mechanical Engineering Dept.
Dhofar University, Salalah,
Sultanate of Oman, Middle East.

Viennese waltz behaviour means combined translation and rotation. This motion is applied to a six-legged omnidirectional robot. Such behaviour is highly useful because it represents generalised omnidirectional motion of a vehicle in a ground plane. The motion is similar to programming a mobile Stewart platform and is complex problem to program into a six-legged robot especially in real-time. This paper presents a plan to achieve such motion based on a sun and planet wheel model.

Keywords: Emergent behaviour, combined rotation and translation, Viennese waltz behaviour, robot omnidirectional motion, continuous computation, biological computation.


QTAR: Quad-Thrust Aerial Robot

Joshua N Portlock and Samuel N Cubero
Curtin University of Technology, Perth, Western Australia

The Quad Thrust Aerial Robot (QTAR) is a low-cost quadrotor Vertical Take-Off and Landing (VTOL) Unmanned Aerial Vehicle (UAV). The onboard attitude estimation and control system allows for safe and intuitive remote piloting for various aerial imaging applications, both indoors and out. This paper presents the design and development of the QTAR, including the attitude estimation using low-cost inertial sensors and employing a novel Jerk adaptive filtering technique.

Medical 1        2.00 - 3.00


Development of an Intelligent Physiotherapy System

S.L. Chen, W.B. Lai, T.H. Lee, K.K. Tan
Dept of Electrical and Computer Engineering, National University of Singapore

This paper focuses on the development of a portable Intelligent Physiotherapy system. As this system makes use of computer control and programming, it is able to provide a higher level of safety, intelligence, convenience, customization, and portability and enhanced human-equipment interaction compared to many existing physiotherapy equipment such as gym machines and weights. It is also able to provide objective feedback of the physiotherapy patient’s routine; this can lead to more effective treatments as the physiotherapist can make use of the feedback in additional to his professional judgment when analyzing the patient’s condition.  An objective test on the ability of the patient to carry out his routine may also be useful to insurance companies in evaluating the extent of the injury suffered by the patient. Outside the realm of medical treatment, the versatility of the Intelligent Physiotherapy system also makes it ideal as an exercise machine for athletes who want to keep track of their training progresses.


A Display Framework for Presenting Information from a Visual Prosthesis

Jason Dowling1 , Wageeh Boles and Anthony Maeder2
1Queensland University of Technology, Brisbane, Australia
2e-Health Research Centre / CSIRO ICT Centre, Brisbane Australia.

In this paper a novel framework for the dynamic presentation of information to a visual prosthesis recipient is presented.  This framework includes the main factors which impact on blind mobility.   These factors include the current context, scene properties, task undertaken, available sensory information, and environmental factors , in addition to human factors (such as level of training) and prosthesis technology  (for example, camera and electrode array technology).

The electrical stimulation of a component of the human visual system can result in the perception of blobs of light (or phosphenes) in totally blind patients.  By stimulating an array of closely aligned electrodes it is possible for a blind person to perceive very low-resolution phosphene images.  Using this approach, a number of international research groups are working toward developing multiple electrode systems (called visual prostheses or Artificial Human Vision (AHV) systems) to provide a phosphene-based substitute for normal human vision.  Computer vision methods provide a critical link between the camera and electrode array of an effective visual prothesis.

Medical 2        3.30 - 5.00


An autonomous surgical robot applied in practice.

P.N.Brett1, R.P.Taylor1, D.Proops2, M.V.Griffiths3 and C.Coulson1
1Aston University, Birmingham UK
2Queen Elizabeth Hospital Birmingham, UK
3Saint Michaels Hospital, Bristol, UK

Robotics in surgery is a field that has been explored over two decades leading to a series of commercially available systems.  The systems developed have demonstrated the application of mechatronics in challenging working conditions in practice.  The aim is to apply the benefits of such machine systems to achieve precision in placing tool points at the position of the target tissue interface.  Safety, sterility, surgical practice, the operating environment and operating protocol are significant driving factors to deriving acceptable solutions.  

Registration between tool path and the tissues of the patient are often achieved by localising patient scan data with the machine. As such surgical robots have not been autonomous and have been working on paths and regions defined from pre-operative scan data. In this paper the application of the first autonomous surgical robot is described applied to a microsurgical task.  The robotic micro-drill was first used in cochleostomy and moves on a linear tool path with respect to tissue-tool interaction encountered and determined from the sensory information.  The departure from the traditional approach of pre-defining motion is necessary at the scale of micro-surgical tool action where tissue deflection in response to applied loads is regarded significant.  Earlier versions of this device have been reported in terms of the sensory approach.  Here the emphasis is towards the application in clinical trials of the final device.  This paper describes the purpose and design with respect to the procedure, surgical protocol, operating room environment, and the performance achieved by the mechatronic device in practice.


Intelligent Approach to Cordblood Collection

S. Chen, K.K. Tan, T.H. Lee, S. Huang and K.Z. Tang
Dept of Electrical and Computer Engineering, National University of Singapore

In this paper, the development of an intelligent umbilical cord blood (UCB) collection system is presented. The hardware architecture, software development platform, user interface, and all constituent control components will be elaborated on in the paper. The overall control system is comprehensive, comprising of various selected control and instrumentation components, integrated within a configuration of hardware architecture centred around a dSPACE DS1002 Digital Signal Processing (DSP) board.


Deformable Object Simulation with Fluid Model

Yongmin Zhong1, Bijan Shirinzadeh1 and Julian Smith2
1Robotics & Mechatronics Research Laboratory,
Department of Mechanical Engineering, Monash University, Australia
2Department of Surgery, Monash Medical Centre,
Monash University, Australia

This paper presents a new methodology for deformable object simulation by drawing an analogy between ideal fluid and elastic deformation. The potential energy stored due to a deformation caused by an external force is calculated and treated as the source injected into the fluid system, as described by the energy conservation law. A fluid based methodology is proposed for soft object deformation. An improved fluid model is developed for propagating the energy generated by the external force among mass points. A method is presented to derive the internal forces from the potential energy distribution. This methodology can not only deal with large-range deformation, but it can also accommodate isotropic, anisotropic and inhomogeneous materials through simply changing the fluid coefficients. Examples are presented to demonstrate the efficiency of the proposed methodology.

Day 2 - Wednesday 6 December

Other Techniques 1         9.00 – 11.30


Sensing Objects on a Belt Conveyor Using an Ultrasonic Rangefinder Array and Neural Net Based Data Processing

Changwoo Ha and Yongtae Do
School of Electronic Engineering, Daegu University, 712-714, Korea

This paper describes the construction and operation of a sensor system using multiple ultrasonic rangefinders to measure the position of a moving object on a belt conveyor. Nine Polaroid transducers are positioned in a 2D matrix array. The output of the sensor system is a sparse range image of five-pixel width. A neural network is employed to improve the image. The feedforward multilayer network is trained automatically using the data collected by a vision system for bar-coded boxes of known sizes. In an experiment of measuring the centroids of objects, the use of a neural network resulted in about 10% accuracy improvement.


A Power and Bandwidth Efficient Remote Wireless Video Surveillance System Using Joint Source-Channel Coding

Wei Xiang
Faculty of Engineering and Surveying
University of Southern Queensland
Toowoomba, QLD 4350, Australia

A novel wireless remote video surveillance system based on joint source-channel coding is proposed in this paper. The proposed system exhibits superior bandwidth and power efficiency thanks to the adoption of the proposed iteratively decodable variable-length codes (IDVLCs). An experimental platform was built and simulation results were shown to demonstrate the superior performance of the proposed video surveillance system in terms of both objective and subjective measurements.


Direct Kinematics of 3DOF Planar Parallel Mechanisms: A Discussion

Denny Oetomo1, Hwee Choo Liaw1, Gursel Alici2, Bijan Shirinzadeh1, and Yongmin Zhong1.
1Department of Mechanical Engineering,
  Monash University, Clayton, VIC 3800, Australia.
2 School of Mechanical, Materials, and Mechatronics Engineering,
  University of Wollongong, NSW 2522, Australia.

This paper presents the direct kinematic solutions of 3DOF planar parallel mechanisms, specifically of the 3RRR mechanism. The direct kinematic problem of the mechanism generally results in a polynomial solution.  To obtain an analytical (closed-form) solution, it is necessary for the polynomial to be of degree 4 or less. This paper presents a special configuration that reduces the polynomial solution to a 4th degree. This special configuration further decouples the direct kinematics of the position and the orientation of the end-effector into two cascaded quadratic equations.  A numerical example is carried out to verify the results of this simplification. The existence of the closed-form solution would increase the accuracy of the direct kinematic solution and improve computational efficiency, as numerical iterative method is not required. This result provides an efficient computational method for a very useful configuration of planar parallel manipulators.


Modelling pneumatic muscles as hydraulic muscles for use as an underwater actuator

Kenneth, K.K.Ku and Robin Bradbeer
Department of Electronic Engineering, City University of Hong Kong, Kowloon, Hong Kong

Pneumatic muscles have been used for a number of years as actuators in robotic systems, usually for those that mimic human actions. They are most commonly used in systems designed to aid physically handicapped people. It has become clear that if the muscles could be powered hydraulically instead of pneumatically, then they would provide a very efficient actuator for underwater use. This paper looks at preliminary results of such an application.


Vision Techniques 1        11.30 - 12.30


Advanced 3D Imaging Technology for Autonomous Manufacturing Systems

A. Pichler, H. Bauer, C. Eberst, C. Heindl, J. Minichberger
PROFACTOR Research, Im Stadtgut A2, 4407 Steyr-Gleink, Austria.

Today’s markets and economies are becoming increasingly volatile and unpredictable; they are changing radically and even the innovation speed is accelerating. Manufacturing and production technology and systems must keep pace with this trend.  To counter these challenging prospects, technologies enabling autonomous production systems are needed. The paper gives an overview of a novel innovative vision system embedded in manufacturing applications. The system is able to handle (1) unknown products, (2) in-available data, (3) large disturbances and deviations (shape, position, or even part type) and (4) even very small lot-sizes. The concept foresees a full integration into production technology. Products to be processed are passed through a flexible 3D sensor cell. In principle, the 3D imaging process is based on subsequent modules: (1) 3D data retrieval, (2) 3D reconstruction and (3) 3D data interpretation. The proposed system adapts its sensor system to arbitrary product geometry and builds a 3D world model of the task space using fast and robust 3D object recognition system. Having a flexible 3D imaging system paired with automatic process planners, the approach has proven to be a promising alternative to conventional teach-in and OLP based programming and is feasible especially for highly variant, low volume parts.    Robust 3D imaging technology contributes to the realization of autonomous manufacturing systems if part or scene complexity is very high.

Keywords:  3D imaging, object recognition, robotic, process planning.


A vision system for depth perception that uses inertial sensing and motion parallax

V Bečanović and XB Wang
Kyushu Institute of Technology, 2-4 Hibikino, Wakamatsu, Kitakyushu, Japan.

A vision system is proposed that calculates the distance to an object based on geometrical relations of perceived object point and observer motion. The system will in the reference frame of the observer give absolute distance estimates to objects that are static or moving with close to constant velocity. The distance algorithm was validated in a ray tracing Monte-Carlo simulation and a corresponding experiment was also successfully performed. In the experiment the perceived object motion was acquired with a contrast enhancing silicon retina type of imager and the motion of the observer was estimated with a low cost inertial sensor. The experiment was performed with the vision system accelerated with the help of gravity and was tested for static objects in a range from 0.5 to 3.0 meters showing promising results.

Education 2         2.00 - 3.00


The Hong Kong Underwater Robot Challenge

Robin Bradbeer
Director, Hoi Ha Wan Marine Science and Engineering Laboratory,
Department of Electronic Engineering, City University of Hong Kong,
Kowloon Tong, Hong Kong

Marine conservation is usually a subject that is taught to life sciences students at secondary schools. As such, those students interested in marine conservation tend to come from biology and related subjects. The project described in this paper aims to bring the concepts of using technology for marine conservation to those students of a more technological background. The great interest in the ocean surrounding Hong Kong led to the idea of having the students build underwater robots, similar to those of the MATE/MTS competition. The parts kit includes motors, wire, propellers and switches, and the robots’ performance is tested in a pool by carrying out tasks under remote control.  This paper presents an analysis of the event as well as the results of taking part in the World Finals in Houston in June 2006.


Project-oriented low cost autonomous underwater vehicle with servo-visual control for Mechatronics Curricula

C. A. Cruz-Villar1,  V. Parra-Vega2, and A. Rodriguez-Angeles1
1Center for Research  and Advanced Studies (CINVESTAV I.P.N.),  Electrical Engineering Department, Mechatronics Group
2CINVESTAV I.P.N., Saltillo, Robotics and Advanced Manufacture Group

This paper describes a  project oriented Autonomous Underwater Vehicle (AUV)  that integrates three different courses in a Mechatronics Master Sciences academic program. The postgraduate program in Mechatronics at CINVESTAV includes the courses:
Real Time Programming, CAD/CAM/CAE, and Modeling and Simulation of Mechatronics Systems. The students  simultaneously pursue these three courses that last 14 weeks in two weekly sessions of two hours each.  It is intended that the students develop a final project per course to apply the acquired knowledge. However during the first term of 2006, it was proposed to integrate the final project, so that it involves goals and tasks of the three courses.
For integration purposes it was proposed to take a radio control system which could be modelled and modified to apply reverse engineering and real time control.  On looking at commercial radio control mobile systems, a micro submarine was selected as it requires a small working area and presents challenging issues such as hydrodynamic effects and under-actuation in some degrees of freedom.

Vision Techniques 2        3.30 – 5.00


Particle size determination in granular images using edge detection, with application to agricultural products

Ronald Elunai and Vinod Chandran
Queensland University of Technology

A method for determination of grain size from granular textures is developed. The method is one which at-tempts to capture the graininess distribution directly from the digital image. It employs classical edge detection and computes the average distance, or pixel runs, between successive edges. With Uniform digital image acquisition conditions, the method can discriminate images with different granularities. Two experiments were conducted, one with subjective ranking of grain images, and another experiment for measurement of absolute graininess. Both experiments reveal the suitability of the method in measuring granularity and produce results that correlate with perceptual experiments.


Vision Based Person Tracking and Following in Unstructured Environments

Mahmoud Tarokh and John Kuo
Department of Computer Science, San Diego State University,
San Diego, CA 92124, USA.  

The paper describes a method for the detection of a person from the environment by a vision system, and tracking and following the person by a robot equipped with a camera.  The person detection is achieved by a technique based on colour and shape recognitions.  Based on the features provided by the visions system, a fuzzy control scheme is proposed for tracking the person.  Four quantities are controlled, namely, camera pan and tilt, and robot speed and steering.  The software consists of several components for image processing, low level motor control and high level fuzzy control.  The results of person following are reported for indoor and outdoor environments using an all-terrain rover on which an articulated camera and a lap top computer are mounted.  These trials include passage through labs and hallways, moving up steep and winding dirt trails, traversal over rocky terrain, and person following with incomplete image and partial occlusion.


Simple, Robust and Accurate Head-Pose Tracking Using a Single Camera

Simon Meers, Koren Ward and Ian Piper
University of Wollongong, Australia.

This paper describes an inexpensive, robust method for tracking the head position and orientation of the user by using a single low-cost USB camera and infrared light emitting diodes concealed within spectacle frames worn by the user. Unlike gaze and head-pose tracking systems which rely on high-resolution stereo cameras and complex image processing hardware and software to find and track facial features on the user, the proposed system is able to efficiently locate and track the head's orientation and distance relative to the camera with little processing. Due to the infrared light emitting diodes having fixed geometry, the system does not have to contend with the varying facial features of different users and therefore does not require any calibration procedure or training to accommodate different users. Furthermore, the proposed system is unaffected by varying lighting conditions and can be used in the dark. Experimental results are provided demonstrating a head-pose tracking accuracy of within 0.5 degrees when the user is within one meter of the camera. This compares well with more expensive head tracking systems.

Day 3 - Thursday 7 December

Vision Applications 1         9.00 - 10.30


Millimetre wave radar visualisation system:  practical approach to transforming mining operations

E. Widzyk-Capehart1, G. Brooker2, S. Scheding2, A. Maclean2, R. Hennessy2, C. Lobsey2 and M. Sivadorai1
1CRCMining, The University of Queensland, Brisbane, Australia
2CRCMining and Australian Centre for Field Robotics, University of Sydney, Sydney, Australia

This paper describes the development of millimetre-wave frequency modulated continuous wave radar and its application in the mining industry. Assessment of various sensors: laser, sonar, visual and microwave radar, against the mm wave radar sensor proves radar performance superiority in difficult mining conditions. The implementation and performance verification of the radar sensor for range and 3D profiling were undertaken in an underground environment for stope fill and ore-pass monitoring and in surface mines for dragline environmental mapping and rope shovel bucket and dig face imaging. The millimetre wave radar visualisation system development, testing, on-site applications and the benefits to the mining industry are hereby discussed.


Minutiae-based Fingerprint Alignment Using Phase Correlation

Weiping Chen and Yongsheng Gao
School of Engineering, Faculty of Engineering and Information Technology,
Griffith University, Australia

Fingerprint alignment is an important step in fingerprint recognition. The performance of fingerprint recognition relies heavily on the accuracy of fingerprint alignment. The minutiae-based approach is the most popular approach in fingerprint alignment. However, most existing minutiae-based approaches are using additional features (ridge pattern and nearest minutiae) or reference points (core and delta points).  These approaches need to search for the best correspondence of pairs to estimate the alignment parameters. The problem of lost minutiae or spurious minutiae always occurs during the minutiae detection process. Hence, the corresponding pairs or reference points may not be found under this condition. This paper proposes a new minutiae-based fingerprint alignment algorithm using phase correlation. First, we convert minutiae sets into 2D image spaces called Minutiae Direction Image (MDI). Then the transformation parameters can be calculated using phase correlation between two MDIs directly. Experimental results show that the proposed approach performed well in aligning fingerprint minutiae sets and greatly improved the economy of storage space.


Machine Vision for Beer Keg Asset Management

Michael Lees1, Duncan Campbell2, Andrew Keir2
1Foster’s Australia, Yatala Operations
2School of Engineering Systems, Queensland University of Technology

 A typical large brewery could have a keg fleet size in the order of hundreds of thousands. For some breweries, the asset value of this fleet is second only to fixed plant. The annual rate of attrition within the fleet can range from 5% to 20%, a sizable figure in dollar terms with a stainless steel keg costing around USD100. There is a business case for a keg asset management system that can help to reduce the annual rate of attrition and supply chain cycle time. Established solutions such as bar codes and RFID tags are costly as they require a modification to every keg. The feasibility of a machine vision tracking system based on the optical character recognition (OCR) of the keg's existing serial number is explored. With prospective implementation in the keg filling line, a process is proposed which is based on neural network OCR. A recognition rate of 97% was achieved for kegs with non-occluded serial numbers with realistic scope for further improvement.

Vision Applications 2         11.00 - 12.30


On-the-go machine vision sensing of cotton plant geometric parameters: first results

Cheryl McCarthy, Nigel Hancock and Steven Raine
Cooperative Research Centre for Irrigation Futures,
National Centre for Engineering in Agriculture
Faculty of Engineering and Surveying, University of Southern Queensland, Toowoomba, Australia.

Plant geometrical parameters such as internode length (i.e. the distance between successive branches on the main stem) indicate water stress in cotton.  This paper describes a machine vision system that has been designed to measure internode length for the purpose of determining real-time cotton plant irrigation requirements.  The imaging system features an enclosure which continuously traverses the crop canopy and forces the flexible upper main stem of individual plants against a glass panel at the front of the enclosure, hence allowing images of the plant to be captured in a fixed object plane.  Subsequent image processing of selected video sequences enabled detection of the main stem in 88% of frames.  However, node detection was subject to a high false detection rate due to leaf edges present in the images.  Manual identification of nodes in the acquired imagery enabled measurement of internode lengths with 3% standard error.


Visual Position Estimation for Automatic Landing of a Tail Tail-Sitter, Vertical Takeoff and Landing Unmanned Air Vehicle

Allen C. Tsai, Peter W. Gibbens, and R. Hugh Stone
School of Aerospace, Mechanical and Mechatronic Engineering, University of Sydney.

Vertical Takeoff and Landing (VTOL) Unmanned Air Vehicles (UAV’s) combine the operational benefit of efficient forward flight with the capability to land vertically on unprepared sites. Vertical landing in small spaces does, however, require very precise relative navigation and guidance. Visual systems promise a very real benefit over systems like GPS, in that the accuracy of the relative position estimates should improve as the vehicle gets closer to the landing target. This paper describes progress toward the development of a visual sensor system to provide information on aircraft attitude and position relative to a designated landing target while hovering on descent.  The system combines principles of target recognition with attitude estimation using perspective projection. These data are fused with inertial attitude information to assist in the identification of the lateral and vertical distances to the target for use in the guidance system.  The paper shows preliminary results of position estimation using image data taken during flight trails.  It compares these results to GPS position estimates and discusses the relative accuracy.


Machine Vision Application for Machined Components Surface Roughness Assessment in the Micro and Nano-Scale Regions

Ghassan A. Al-Kindi1, Bijan Shirinzadeh2 and Yongmin Zhong2
1Department of Mechanical Engineering, University of Technology, Baghdad, Iraq
2Robotics & Mechatronics Research Laboratory, Department of Mechanical Engineering, Clayton, VIC 3800, Australia

This paper presents a vision-based approach for valid assessment of surface roughness in both micro-scale and nano-scale regions. To enable data comparisons, three sets of surface data in the micro-scale and nano-scale regions are acquired by using a CCD camera, a video-based optical microscope and a stylus instrument. Data filtering and analysis procedures are applied to the acquired data. Results for computation of roughness parameters by using vision data provide adequate values for assessment of surface roughness in the same manner as the stylus based technique. No obvious changes in the computed roughness parameter values are resulted from the micro-scale and nano-scale regions. In the nano-scale region, a cavity graph technique provides distinguishable forms of graphs, which tend to more gradual increase of the cavity percentage to denote the collection of the macro surface details. In addition, an auto correlation technique applied in the nano-scale region succeeds to discriminate the surface irregularity relationships with respect to their periodicity and randomness. The overall acquired results indicate that vision systems are a valid source of data for reliable surface roughness evaluation in both micro/nano-scale regions. The results are very useful in achieving commercial 3D vision based micro/nano-scale roughness measurement systems for industrial applications.

Other Techniques 2         2.00 - 3.00


Non-linear Observer for Slip Estimation of Tracked Vehicles

Z.B. Song1, Y.H. Zweiri2, K. Althoefer1 and L.D. Seneviratne1
1Department of Mechanical Engineering,
King's College London,
Strand, London WC2R 2LS, UK.
Telephone: +44 (0)207 848 2236
2Department of Mechanical Engineering,
University of Mu'tah,
Karak, Jordan

Slip plays an important role in determining tracked vehicle tractive forces. Thus accurate estimation of slip is essential in developing autonomous navigation strategies for unmanned ground vehicles operating in unprepared terrain.

This paper presents an accurate and robust technique for the estimation of tracked vehicle slip parameters; the slips of outer and inner tracks and the slip angle. The technique uses a sliding mode observer (SMO) with sprocket wheel angular velocities and vehicle trajectory as inputs, and estimates slip parameters by minimizing the errors between the predicted trajectory and the measured trajectory. The SMO is based on a full kinematic model of the tracked vehicle and a Utkin SMO. The conditions for stability of the SMO are derived using Lyapunov stability theory. The experimental results show that the slip estimate using the SMO can accurately capture the trend of measured slip parameters.


Mobile Robot Navigation using Alpha Level Fuzzy Logic System: Theoretical and Experimental Investigations

S.Parasuraman1 , V.Ganapathy1, Bijan Shirinzadeh2, Yongmin Zhong2
1Monash University, Malaysia, 2Monash University, Australia.

In this paper, behavior rule selection or decision-making procedure incorporating Alpha-level fuzzy logic system is discussed and used for selecting an appropriate action during mobile robot navigation. In this methodology, environmental perceptual field is well defined by a set of control parameters through alpha level intervals of fuzzy sets and their fuzzy membership value. The expected output control parameters are also partitioned by alpha level intervals. In the proposed method, the alpha level threshold maximizes the truth for a particular behaviour rule that needs to be fired at a time. This methodology presents the mathematical aspects of resolving conflicts when more than one context rule of the same kind is in action. This approach provides a formal methodology for representing and implementing the human expert heuristic knowledge and perception-based action in mobile robot navigation. In this approach, the operational strategies of the human expert driver are transferred via Alpha level fuzzy logic to the robot navigation in the form of a set of simple conditional statements composed of linguistic variables. The proposed behavior rule selection method is validated by real world experiments using active media pioneer robot and the results are discussed and compared with most accepted methods.

Short video   3.30


More machine vision applications in the NCEA

John Billingsley
University of Southern Queensland

This is a second showing of a video presentation prepared for the ICMT conference in Mexico.
With the trend of computers towards convergence with multimedia entertainment, tools for vision processing are becoming commonplace. This has led to the pursuit of a host of unusual applications in the National Centre for Engineering in Agriculture, in addition to work on vision guidance. These range from the identification of animal species, through the location of macadamia nuts as they are harvested and visual tracking for behaviour analysis of small marsupials to the measurement of the volume of dingo teeth.
Keywords: Agricultural applications, Machine
vision, Shape matching, s-psi.

Vision tutorial    3.45  Mark Dunn, NCEA