The first batch of Mechatronics Students will be graduating this semester. The graduating students have opted for both Thesis and Projects. This newsletter comprises of a brief description of their projects.

Project Title:

Two Link Manipulator Arm Positioning Using Neural Network, Sliding Mode, and Nonlinear Controllers

Brief Description:

The objective of the proposed project is to control the position of a two link manipulator arm using different control techniques such as neural network, sliding mode, and nonlinear controllers. This project will cover designing and building of a two link manipulator arm, then the design and implementation of real time controllers that well enable the control of the end effecter of the manipulator arm.

Hardware part of the project will involve designing and building a light weight two link robotic manipulator arm and interfacing it to a personal computer where the control algorithms will be implemented.

Model-based and model-free control schemes will be considered on this project. First the mathematical model of the robotic arm will be derived and verified experimentally. Model-based nonlinear and sliding mode robust tracking controllers that are selected form literature will be designed and implemented to control the position of the two link manipulator utilizing the derived mathematical model, then real time testing of the tracking and positioning performance of these controllers will be carried out.

Neural network controller will also be used to control the robotic arm. Real time links responses will be experimentally collected and used to train the neural network controller. This network will then be implemented real time to control the positioning and tracking of the two link manipulator.

The real time tracking and positioning performance of the proposed controllers will be compared. Controllers' algorithms and implementation complexity and required torque by each controller also will be compared.

A photo of the setup.

Student :

Muaatasem Awda

Project Advisors:

Dr. Mohammad Jarrah.

Dr. Yousef Al-Assaf.

____________________________________________________________________

Project Title:

Study Intersection Collision Avoidance System Deployment for a Modern City.

Brief Description:

The expansion of computer and sensing technologies into the roadside and vehicular environment and the corresponding drop in prices coupled with an increase in accuracy are starting to make the detection of speed, direction, and position of a vehicle more cost effective. This makes it possible to use this information to provide warnings to vehicle operators of possible traffic conditions that may result in a collision.

The use of video tracking system, millimeter Doppler radar, Differential Global Positioning System (DGPS), Geographical Information Systems (GIS), and digital map capabilities in a coordinated system will enable the tracking of vehicles traveling through intersections with an accuracy of one meter or less. The use of environmental sensors will enable the real-time determination of roadway surface conditions: dry, wet, or icy. The use of radio technology can enable data exchange between vehicles and between vehicles and the roadway infrastructure to improve the safety of the streets and highways.

An Intersection Collision Avoidance (ICA) system, which integrates vehicle tracking, roadway surface monitoring, and communication systems, will be able to utilize the information attained from its subsystems and determine whether a vehicle is on the course toward a potential collision. It is anticipated that the ICA system will be capable of monitoring vehicular traffic flow through the intersection, predicting the possibility of signal violations, delivering warning messages to targeted devices, and communicating time-critical messages in an appropriate radio band to suitably equipped vehicles.

The implementation of an ICA system can follow two totally different strategies, which can be describes as intersection-based and vehicle-based systems. Under an intersection-based deployment scenario, the Intersection Subsystem will be responsible for collecting real-time information from a variety of subsystems, evaluating potential collision threats, and then implementing appropriate actions to avoid potential collisions. Each approaching vehicle simply reacts to the messages disseminated from the Intersection Subsystem. On the other hand, vehicles in a vehicle-based system are responsible for making their own decision to minimize potential collision threats based on the prevailing information broadcasting from the Intersection Subsystem as well as information gathered from their on-board devices. In the vehicle-based system operation, the Intersection Subsystem acts as an information resource for all properly equipped approaching vehicles.

A diagram representing an ICA system.

Student :

Ahmad Al Zarooni.

Project Advisor:

Dr. Mohammad Jarrah.

____________________________________________________________________

Project Title:

Two Link Manipulator Arm Positioning Using Model-Free Controllers

Brief Description:

In this project model-free controllers are proposed to accomplish the task of positioning a
two links manipulator arm using neural networks, neuro-fuzzy, and polynomial classifier.
The objective of the proposed project is to compare between the performances of each of
the above mentioned controller in achieving the required task. The motivation of this
project is to experience the feasibility of using the latest technologies and techniques to
handle and control a real life application.

Each and every one of the proposed controllers will be developed and applied separately
to control the position of a planner two links manipulator arm which was built by me and
my colleagues last semester. A comparison will be done on certain control properties of
the different applied control schemes.

The two link manipulator arm.

Student :

Rabih El Assadi

Project Advisors:

Dr. Mohammad Jarrah.

____________________________________________________________________

Thesis Title:

Model Identification of Unsteady Aerodynamic Loading of Delta Wings for High Angle of Attack Using a Polynomial Networks

Brief Description:

A novel method based on a polynomial networks is proposed to model force and moment data obtained from forced oscillation tests of Delta wing at high angles of attack.
Several examples are considered to assist the prediction capabilities of the proposed model as compared to other parameter identification techniques. These include dynamic wind tunnel test data for NASA X-31 and F-16 for large amplitude forced oscillation data of a 70-deg delta wing.

Different techniques will be discussed for the modelling of force and moment coefficients, based on the previous work, to show the efficiency of the polynomial modelling technique.

F-16C at 60 degrees angle of attack. Surface cuts on top view are of vorticity magnitude. Red streamlines represent lex vortices and yellow streamlines represent forebody vortices.

Student:

Mohammad Al-Khedher.

Thesis Advisors:

Dr. Khaled Assaleh.

Dr. Mohammad Jarrah.

____________________________________________________________________

Thesis Title:

Automatic classification in the field of marble.

Brief Description:

The main goal of this research is to implement an automatic system to classify marble tiles. Different image processing and pattern recognition techniques have been implemented. The features from images were extracted using different feature extraction methods. The extracted features were used for training the Classifier. Two classifiers were evaluated, feed forward neural network and polynomial classifiers.

The main goal of this research is to implement an automatic system to classify marble tiles.
Marble texture is a demanding and challenging task, because the texture is often non homogenous. There are many kinds of marble tiles; we roughly classify tiles as basic tiles and their subclasses. Basic types have similar characteristics like color and pattern on them .Subclass have only slight differences from each other such as the percentage of color, length and thickness f the veins. The samples chosen for testing the system have been Italian marble tiles of type "Carrera ". This basic type of marble coming with white, pink or silver background, brown, green, red or black veins.

Different image processing and pattern recognition techniques have been studied extensively in the process. The features from images were extracted using different feature extraction methods. Statistical methods represented by the most popular one co-occurrence matrices were compared with frequency transforms methods. Transform methods based on frequencies like discrete cosine transform (DCT) and Wavelets proved their strength in packing the energy of images more than statistical ones. Hybrid methods as combination between the above methods were also considered.

The extracted features were used for training the Classifier. Two classifiers were evaluated, feed forward neural network and polynomial classifiers. Polynomial classifiers improved the
recognition rate with the advantage of a decrease of computation time and memory storage.

Marble Samples.

Student:

Sufian Al Titi.

Thesis Advisors:

Dr. Yousef Al-Assaf.

Dr. Khaled Assaleh.

____________________________________________________________________

Thesis Title:

Next Generation 10 Gb/s RZ Optical Transmitters.

Brief Description:

The objective of this research is to propose a new chirped RZ transmitter to provide telecom carriers with a low cost solution for upgrading the existing 2.5 Gb/s fiber links to 10 Gb/s. Computer simulation results indicate that transmission up to 2500 km is possible using the chirped RZ source but not with a chirp free RZ source nor an NRZ source.

Many next generation optical transmission systems at 10 Gb/s will be utilizing RZ line coding to mitigate fiber dispersion and nonlinear effects for extending the reach of fiber links to a few thousand kilometers without electronic regeneration. This thesis will introduce a chirped RZ transmitter at 10 Gb/s. A computer simulation will be used to demonstrate the advantages of this transmitter compared with the chirp free RZ and NRZ transmitters. The proposed chirped RZ transmitter provides telecom carriers with a low cost solution for upgrading the existing 2.5 Gb/s fiber links to 10 Gb/s without reinserting Dispersion Compensating Fiber (DCF) modules at all EDFA (Eribium-Doped Fiber Amplifier) sites in the link but using only DCF modules at either of the receiver/transmitter sites.

The new RZ Transmitter.

Student:

Amira Al Houli.

Thesis Advisor:

Dr. Aly Elrefaie.

____________________________________________________________________

Thesis Title:

Brain Computer Interface based Prosthesis Control.

Brief Description:

A Brain Computer Interface System will be designed and implemented which collects brain EEG signals, determines the required movement and controls the movement of a robotic arm in the required direction. The work which will go into my Mechatronics Engineering thesis will potentially benefit members of several groups. The BCI system will provide aid to persons with impaired muscle or movement control. The system, the method and the data could be of value to neurophysiologists, military applications, and could shed light on the complex operations of the brain. Furthermore, the system could contribute to the collective effort of scientists and engineers currently developing BCI systems.

Currently, the most active area of research with respect to BCI systems is the non-invasive BCI system approach. Non-invasive BCIs work by obtaining EEG readings from electrodes placed on the scalp, under it, or on the brain's surface. Due to safety concerns, the proposed BCI will use electrodes placed on the scalp. The EEG signals obtained from these electrodes are the input to the BCI system.

The non-invasive signal activity that is currently being researched can be classified as follows:

• EEG Pattern Mapping.
• Slow Cortical Potentials.
• Evoked Potentials
  • Visual Evoked Potentials.
  • Steady State Visual Evoked Potentials.
  • P300 Detection.

Brain EEG signals contain vital information about the functions being carried out by the brain. The brain functions are concentrated in specialised areas of the brain. For example, when a person imagines limb movement, the EEG signals obtained from the sensorimotor cortex indicate this thought process.

Furthermore, certain signal patterns can even indicate the direction of the imagined motion. This is the essence of the thesis. A pattern classifier will be developed to identify which limb is being imagined, and the direction is its imagined movement. The type of signal used will determine the pattern classification techniques. SSVEP and EEG pattern mapping techniques will be used to implement the proposed BCI system.

The Graphical User Interface of the BCI system.

Student:

Hania Rana.

Thesis Advisors:

Dr. Hasan Al-Nashash.

Dr. Yousef Al-Assaf.

____________________________________________________________________

Thesis Title:

Improving GPS accuracy using neuro-fuzzy systems

Brief Description:

Single GPS receiver is capable of horizontal accuracy within 10 to 20 meters. Manipulating a low cost GPS receiver positioning data with artificial intelligence methods may improve the accuracy up to an acceptable value acceptable good enough for such applications. In this research, neuro-fuzzy systems will be considered to improve the accuracy of GPS receiver. Various GPS parameters are used to develop these systems. The developed methods are tested on static and dynamic positioning.

DGPS and RTK GPS systems give accurate positioning but their use is not justifiable in applications that do not need their accuracy. On the other hand stand alone GPS may not give the required accuracy in such applications.

Several studies and research had been carried out to improve the quality (accuracy) of a low cost GPS (stand-alone mode) using fuzzy logic and neural network. These methods have utilized GPS satellite parameters to improve accuracy. The two parameters used are the Position Dilution Of Precision (PDOP) and Signal-to-Noise Ratio (SNR).

DOP is a dimensionless number that accounts for the contribution of relative satellite geometry to errors in position determination. DOP has a multiplicative effect on the UERE. Generally, the wider the spacing between the satellites being tracked by a GPS receiver, the smaller the position error. The most common quantification of DOP is through the position dilution of precision (PDOP) parameter. PDOP is the number that, when multiplied by the root mean square (rms) UERE, gives the rms position error. Other DOPs include the geometric dilution of precision (GDOP), horizontal dilution of precision (HDOP), vertical dilution of precision (VDOP) and time dilution of precision (TDOP).

The objective of the proposed research is to develop intelligent systems that can be associated with stand alone GPS systems to improve their accuracy to the point acceptable by many engineering applications. The methods to be developed make use of satellites parameters provided by the GPS system

In comparison to what has been done in previous research the proposed research will address the following developments:

1- Develop other structures and intelligent systems to test their capabilities in improving GPS positioning accuracy.

2- Include other critical satellites parameters in the development of such intelligent systems. In addition to the PDOP and SNR we will utilize the User Equivalent Range Error (UERE).The UERE error is the equivalent error in the range between a GPS receiver and a satellite. UERE errors originate from different sources and thus are independent of each other. The total UERE is the square root of the sum of the squares of the individual errors. A prediction of maximum anticipated total UERE (minus ionospheric error) is provided in each satellite’s navigation message as the user range accuracy (URA). We believe this parameter is critical indicator of the error and would improve accuracy if included.

3- Previous research worked on static position accuracy. This work will consider dynamic applications. Such applications give rise to many other considerations and challenges in GPS correction. Total correction on the line of movement is important and the correction time is limited and is a factor in the correction speed.

4- In this work the results of the other research for static positioning will be reproduced and compared to the results obtained by our developed methods.

The GPS postioning system

Student:

Samer Imad Akoum

Thesis Advisors:

Dr. Yousef Al-Assaf.

____________________________________________________________________

Thesis Title:

Fuzzy Autonomous Flight controller.

Brief Description:

Unmanned air vehicles are nowadays seen as an area of great importance in the aerospace industry. The most important part in designing an automatic flight control system is the control algorithm that is used in the aircraft, like PID, LQR, state feedback, and other control schemes. The controller has to be able to drive the nonlinear aerodynamic system with maximum stability and fast response.

A comparison between different control methods has been studied and it is decided to use a nonlinear method of control to help adapt the high nonlinearity of our system. The high nonlinearity is due to the coupling between the nonlinear dynamic aircraft model and the nonlinear aerodynamics inputs.

Therefore a digital flight controller using fuzzy logic will be implemented in Motorola MC68HC12 microcontroller. By the end of this thesis we will have a complete auto pilot system for the Piper RC plane that is available in the lab. There will be several phases to accomplish the final goal, studying the UAV aerodynamics, simulating a good Fuzzy Logic Controller, sensor interfacing, integrating all the previous components. Some of these phases are in progress. The interface between several sensors and the microcontroller has already been developed. Most of the sensors are communicating successfully with the MCU like pressure sensor, GPS, magnetic sensor, etc.

The simulation of the linearized system shows that fuzzy controller have a better performance than the LQR controller. The fuzzy controller reduces the transient time and it would decrease the control effort. This shows that a nonlinear controller like FLC can sometimes perform better than the conventional controllers when it is applied to a nonlinear system.

The airplane used for the Autonomous Flight Controller system.

Student:

Mahmoud Hadi.

Thesis Advisors:

Dr. Mohammad Jarrah.

____________________________________________________________________