University of Malta
 

2013 Projects
UOM Main Page
 
 
 
Apply - Admissions 2016
Newspoint
Campus Map button
Facebook
Twitter

Final Year Projects Academic Year 2012/2013

Design and development of an Electronic Fuel Injection system

Student: Anton Tabone
Supervisor: Prof. Carmel Pulé
Co-Supervisor: Ing. Paul P. Debono

2013_A

Introduction
The principal technological innovation of these last 30 years which has had great success in the car industry was the electronic fuel injection system (EFI). Automobiles have improved in several aspects, such as improved reliability, better driveability, reduced CO2 emissions and improved fuel economy [1], [2].

Project Objectives
The aim of this project was to study the operation of a four-stroke internal combustion engine and analyze the fuel and air mixture of the engine which works on a carburettor system. Then, the study moves on how the EFI system works and how to implement the system on an internal combustion engine instead of using its standard carburettor system.

Project Methodologies
To implement the EFI system on a 1300cc manual engine a system had to be designed. The system consists of three sections: the fuel delivery system, the electronic control system and the air induction system.
The fuel delivery system consists of equipment required to supply the fuel to the injector at the required fuel pressure. The air induction system is the section which controls the amount of air which flows inside the intake manifold through the throttle body. The electronic control system consists of a circuitry which controls both, the timing of the injector and the fuel metering [3].

  
Results and Achievements
To convert an internal combustion engine to operate on fuel injection, the maximum fuel flow rate was calculated to select the appropriate fuel injector/s to be able to achieve the peak power attained with its standard carburettor system before conversion.
Several fuel demand values where calculated for different load levels of the engine to achieve a knowhow regarding the variance of the duty cycle of the fuel injector. Then, the pulse width values were obtained depending on the speed of the engine. All the pulse width values were store in a two-dimensional array according to the Load% and the RPM of the engine. The array which stores the pre-calculated data was uploaded into a microcontroller unit which runs the EFI system.  The microcontroller observes continuously the Load% and the RPM of the engine with the use of several sensors around the engine and selects the pulse width value from the array to achieve the best power output performance with the least fuel consumption and the lowest CO2 emissions.

References
[1] J. Hartman, How to Tune and Modify Engine Management Systems, Minneapolis, MBI Publishing Company, 2003.  
[2] J.B. Heywood, Internal Combustion Engines Fundamentals, Singapore, McGraw-Hill Book Company 1988.
[3] Toyota Motor Sales. (2013). Electronic Fuel Injection Overview. [on-line]. Available: http://www.autoshop101.com/forms/h20.pdf

   
   

An FPGA-Based Object Tracking System

Student: Mark Calleja
Supervisor: Ing. Kenneth Chircop

2013_B

Introduction
The use of FPGAs in image processing algorithms is becoming more popular especially where repetitive tasks are performed. Working at much lower frequencies than microprocessors, processing time is still highly reduced due to easily parallelizable processes. 
    
Project Objectives
The aim of this project was to implement a simple object tracking system. This should be able to track objects against a relatively simple background. A typical application would be to track the moon at night. Implemented on a larger scale, such systems could be useful for astronomical studies. 
    
Project Methodologies
An object tracking system needs to be able to get data about the scene, find the object within the scene and move the input sensor in the direction of the object to try and keep it centred. Thus the system was divided into three stages: Input, Image Processing and Output. In the Input stage, an image of the scene is captured using a camera. Therefore an interface was built to set the desired image parameters and to capture the data that is sent by the camera. From this stage, a gray scale image is obtained. This is then used in the Image Processing stage, where the first step is to convert the image from gray scale to binary by comparing all pixels to a threshold value. A binary image is one where the object pixels are represented as 1s while the rest of the pixels are 0s. However, an image typically contains noise which is not desired, as false results might be produced. Therefore the image goes through four stages of filtering in which the object is kept whilst removing the rest of the scene. The last step in this stage is to find the coordinates of the centre of gravity of the object. The output stage consists of a pan-tilt system which moves using two servo motors. The motors are controlled using two PWM signals.  Therefore, depending on the coordinates of the object found in the previous stage, the pulse width of the servo signals is de/increased to turn the servos accordingly. Effectively, by mounting the camera on the pan-tilt system, the object is kept at the centre of field of view of the camera.  

Results and Achievements
Some modification to the original plan had to be made. Unfortunately, the camera interface did not entirely work and so the image could not be captured. Therefore this was replaced by a webcam and a laptop. Using Matlab, the images are captured, converted to gray scale and sent to the FPGA through a UART interface. UART is a serial interface and hence much slower than the original parallel interface. This slowed down the whole system and so some modifications had to be made to compensate. However the rest of the system worked well and with the new input stage, the system is capable of tracking slow moving objects with relatively simple backgrounds.     


 

Remote Monitoring and Control from a Smartphone or Tablet using a Raspberry Pi

Student: Matthias Fenech
Supervisor: Dr. Ing. Andrew Sammut

2013_C

Introduction
A network control system (NCS) was used to monitor and control a system from a remote location. A Raspberry Pi is considered ideal for the server-side of the system due to its low-cost, small design while also offering a great deal of flexibility with regards to its communication links. The client-side is made up of three separate applications, where each application controls a mobile robot in a unique way.

Project Objectives
The primary objective is to control and monitor the system from a remote site while utilizing a Raspberry Pi in the most efficient way possible. These milestones were planned:
Research on server design, video transmission and serial transfer respectively with regards to the Raspberry Pi.
The selection of a webcam and server packages and their implementations.
The setup of the front end and back end of the server.
Experimental implementation and testing of the separate applications.

Project Methodologies
Each application developed implemented a different control method while also presenting a different interface. The client-side provides the user with the information needed to control the system while the server-side script allows the browser to communicate with the Raspberry Pi by using the WebSocket protocol, which in turn communicates serially with a microcontroller located on the mobile robot. All applications developed are optimized for both mobile and desktop browsers, thus able to handle both touch and mouse events.
The three applications developed are: a manual control application that utilizes velocity control based on a Proportional-Integral-Derivative (PID) controller, a virtual line tracking application that operates position control also based on a PID controller and an object following application that uses data received from the camera to follow certain pre-defined objects.

Results and Achievements
The Manual Control application was tested by setting the target velocity of the robot, while comparing this value to the actual speed of the robot. The results obtained as depicted in Figure 1, are then utilized in the tuning of the controller.
The Virtual Line Tracking application was tested by comparing the actual path the mobile robot takes with the one requested. Several different paths were drawn on the browser, comparing the drawn path to the actual path and the odometric reading path.
Figure 2 shows a subset of the mentioned test were the mobile robot was requested to move in an L-shaped fashion.
The Object Following application was tested by fixing the object to be followed onto a uniform background while altering the mobile robot’s pose. The robot then moves to centre the object in the image, resulting in the final horizontal pixel distance from the centre of the object to the centre of the image.

 

Low Level Signal Acquisition for the Life Sciences 

Student: Savio Galea
Supervisor: Ing. Marc A. Azzopardi

2013_D

Introduction
Analytical chemistry is a discipline which involves several analytical methods in order to analyse the chemical composition, identification and measurement of a substance. The use of its knowledge helps in solving science related problems, including assisting physicists in diagnosing diseases, ensuring good quality and safe water and pharmaceutical supplies. Most methods make use of the electrochemical cell made up of electrodes which are immersed in an electrolyte. Different electrochemical techniques are applied to a potentiostat connected to the electrochemical cell. The main objectives of the potentiostat are to control the potential difference and provide the required current to these electrodes.
The advances in these electrochemical techniques led to the introduction of small electrodes such as ultra-microelectrodes. These electrodes imply small currents. The size of these currents means that externally coupled noise has a significant effect on the results attained and hence noise must be mitigated.

Project Objectives
The scope of this project is to analyse all the possible sources of noise found in the potentiostat. A list of all the feasible noise mitigation techniques must be applied to the potentiostat circuit. Another important objective is to implement a computerized, signal generation and data acquisition system using optical transmission. The whole system design as shown in Figure 1, comprising all the noise suppression techniques should be implemented on a multilayer PCB.

Project Methodologies
The first phase involves listing all the possible noise sources found in a potentiostat circuit.
In the second phase the potentiostat circuit must be built and tested. A review on the noise coupled in the potentiostat circuit should be carried out.
The next phase involves an assessment on the noise mitigation techniques which can be applied to the potentiostat. A computerized, signal generation and data acquisition system using optical transmission should also be designed and tested. The analysis on the choice of components should be carried out right afterwards.
The fourth phase should include a careful design of a multilayer printed circuit board layout including all the viable noise suppression techniques
The last phase includes the soldering of components onto the manufactured PCB including the testing and calibration of the system.

Results and Achievements
The potentiostat was first built on a PCB and tested by using an electro-analytical method. The results obtained conformed to theory. After listing all the feasible noise suppression techniques, a signal generation and data acquisition system was implemented by the use of an audio codec, capable of receiving and transmitting digital data in the S/PDIF standard to a computer via TOSlink fibre optic cables. After this arrangement was tested, it showed good results. The last objective of this project was to implement all the system on a multilayer PCB. The PCB layout was designed carefully also including the valid noise suppression techniques. After the PCB was manufactured, all the components were assembled onto the board, including the final testing and calibration of the whole system.


Smart Street Lighting Management System

Student: Warren Gauci
Supervisor: Mr. Paul Zammit
Co-Supervisor: Ing. Evan Dimech 

2013_E

Introduction
In an age where the importance of a clean environment has been recognised while fossil fuel is becoming even scarcer, reducing the consumption of fossil fuel and consequently the emission of green-house gases has become a major goal for most countries around the world. Research has focused primarily on the generation of power from renewable energy sources. A complimentary approach to this problem is to reduce the demand by increasing the efficiency of the current infrastructure. This approach entitles the employment of more established and relatively simpler technologies; therefore, the desired effect can possibly be achieved faster and at a lower expense. In this work, the local street lighting system has been taken as a case study in order to investigate these claims.

Project Objectives

The aim of this project was to design an embedded system that controls the operation of street lighting, in order to obtain a smart system, which reduces the consumption of power and thus operational costs. The objectives were as follows:
Investigation of street lighting requirements as per local and European lighting standards.
Establishment of a management strategy for street lighting that meets the previously mentioned standards as efficiently as possible.
Design of the hardware and software needed to develop an embedded system to manage street lighting according to the established management strategy.
Evaluation of the energy saving potential of the designed and implemented embedded system.

Project Methodologies
The first project phase involved conducting a literature review in order to analyse lighting design considerations, applicable standard regulations, smart street lighting categories and implemented pilot project systems.  It could then be concluded that a fully adaptive type of system would best suit a local implementation. A system layout and a lighting strategy that the latter system should perform in order to achieve the best energy saving results were suggested.
The next phase of the project was the embedded system design phase. Subsequent to a detailed requirement analysis the hardware and software needed to realise the mentioned lighting strategy was designed and rigorously tested. This led to the manufacturing of a final prototype board, shown in Figure 1. An algorithm that interfaces this board with a PC was programmed and the designed strategy was simulated on the PC. Limitations of the designed street lighting strategy were addressed and future recommendations were made.

Results and Achievements
The achieved results include the operational hardware and software that make up an embedded system which performs the function of a lamp controller in a proposed local smart street lighting system. The manufactured embedded system is capable of: measuring ambient light, monitoring the energy consumed by the lamp, detecting a tilted luminaire pole and performing dimming of the lamp. By the use of a ZigBee module it is also capable of wirelessly connecting with a PC and adopting an algorithm that imposes a lighting strategy which was designed to control the amount of light emitted by the lamp, thus reducing power consumption and maximising efficiency.

References
 [1] [P.Van Tichelen, B.Jansen, A.Vercalsteren], ‘[Public Street Lighting]’ [Final Report], 20[07], Vol. [40], No. [56457], pp.[17-20]


A Distance Measuring Technique for Indoor Localisation 

Student: Daryl Martinelli
Supervisor: Ing. Brian Zammit
Co-Supervisor: Dr. Ing. Andrew Sammut

2013_F

Introduction
The use of robotics in everyday life is constantly on the increase. In many instances, an increasing demand for autonomy was experienced, with robots required to operate for long periods of time without any or little human intervention. In order to achieve this high level of autonomy, a navigation function is required. The accuracy and success of the intended operation is only warranted if high localisation accuracy is possible. For outdoor applications a commonly used technology is the Global Positioning System but this is useless for indoor localisation, hence an alternative solution needs to be adopted. Some authors claim that, mobile robot localisation is the “most fundamental problem to providing robots truly autonomous capabilities" [1].

Project Objectives
This work is intended to provide a good insight on a chosen hardware to achieve accurate distance measurements. One technique is then selected and an in-depth analysis of its performance is carried out. The parameters of interest are the execution speed, measurement accuracy, development cost and ease of implementation. The work then focuses on improving the accuracy of the selected technique using signal processing methods.

Project Methodologies
A survey of various measuring techniques that can be adopted for indoor localisation was carried out. The use of ultrasonic technology was pursued for implementation of the distance measurement hardware. The ultrasonic hardware was tested for different characteristics especially, power radiation and detection angles. Figure 1 shows a polar plot for a single transducer power loss against detection angle.
A complete system was set up for ultrasonic distance measuring. This setup uses a time of flight measurement for the ultrasonic signal to propagate in air from point A to point B, to obtain a distance measurement using the speed of sound as a conversion factor.  Distance measuring was obtained using three different methods. A common ultrasonic distance measuring technique was used as a base test, compromising of triggering the reception of an ultrasonic signal based on an accumulated voltage received. Two other methods involving line fitting techniques were implemented to improve accuracy over the basic distance measuring technique. This was done by using the line fitting to back-track the accumulated voltage received to the exact instant when the first ultrasonic signal was received. 

Results and Achievements
The results obtained when using the common method for ultrasonic distance measurement revealed an expected error that increases with respect to distance. The two line fitting methods offer an improved distance measurement, where the errors obtained are less than those obtained with base method, and the most beneficial factor being that the errors do not increase with distance. This is an important result, as indoor localisation can now be accurately obtained and robot autonomy capabilities are increased. 

References
[1] I. J. Cox Blanche, ‘Position estimation for an autonomous robot vehicle’ Autonomous Mobile Robots: Control, Planning, and Architecture, 1991, Vol. 2, pp.285-292. 



Data Fusion of Multi Sensory Information for Reliable State Estimation

Student: Merlin Mifsud
Supervisor: Dr. Ing. Andrew Sammut

2013_G

Introduction
Nowadays most of regular vehicles are equipped with a Global Position System (GPS) that constantly updates the vehicle’s position in order to provide the driver information about his position, with accuracy up to several metres [1]. However, GPS suffers significant limitations such as the slow update rate, particularly at high speeds, and the sensitivity in low signal areas. On the other hand, Inertial Navigation System (INS) provides an alternative, self-contained solution that can provide high update rates, typically up to 100Hz. However, as INS make use of dead-reckoning, low frequency noise and sensor biases are amplified when integrated. Such opposing errors make it possible for both systems to be integrated and provide corrections to each other, which is the topic of this thesis.

Project Objectives

This project aims to obtain a satisfactory level of navigation performance in three-dimensions suitable for autonomous navigation of ground, aquatic or airborne vehicles. Such accuracy can be attained by fusing two independent navigation systems, typically having independent error characteristics.

Project Methodologies

In this project an approach of integrating a GPS and a strap-down INS is performed. The continuous time INS navigation and error equations are achieved. The navigation equation continuously integrates accelerometer and gyro measurements to output heading, velocity and position. These are corrected to the Earth’s rotation, while the vertical channel being furthermore integrated with a Barometer. Inertial Measurement Unit (IMU) error dynamics are observed and plots are attained to justify the unbounded error obtained due to integration. Position fix and velocities are derived from the GPS which are then aided with the INS. An Extended Kalman Filter (EKF) with closed loop integration between the GPS and INS is applied to acquire velocity and position correction, providing closed loop feedback to both navigation systems.  Real-time navigation data is gained and stored, which is then applied off-line using MATLAB to test the integration algorithm.

Results and Achievements

A GPS aided INS was integrated using three methods, uncoupled, loosely coupled using EKF and Standard Kalman Filter (SKF). From Figure 1 it can be appreciated that the EKF attempts a correction at the exact point of turn. As corrections are applied on the error between the current corrected INS state and the current GPS state, corrections are performed to limit the error in relation to the GPS state. This is because the GPS is the long term navigation technique. Therefore, as expected, as the navigation continued the EKF coupling method converged to the GPS position. Conversely, the standard Kalman filter provides corrections in relation to current position and velocities according to covariance values for both INS and GPS independently. By tuning the Kalman gain through covariance values, more prominence was given to the GPS speed, INS orientation and corrected INS position. This combination was chosen in order to counteract for the hardware limitation, which better results were attained as can be seen in the Figure above.

References
[1] A. Schumacher, “Integration of a GPS aided strapdown inertial navigation system for land vehicles”, M.S. thesis, Royal Institute of Technology, Stockholm, Sweden, Mar. 2006, XR-EE-SB 2006:006.


Detection of Critical Driving Situations in Motorcycles

Student: Stefan Sant
Supervisor: Prof. Carmel Pulé

2013_H

Introduction

Motorcycles are one of the most affordable and popular means of transport across the world. However, over the last fifteen years, the portion of motorcycle crash fatalities has risen. Motorcycle and moped fatalities, often referred to as Powered Two Wheelers (PTW), accounted for 16% of the total number of road accident fatalities in 2009 in the EU 24 countries [1].
An analysis of an accident database and a motorcycle rider survey [2] revealed the following as causes for accidents that could be potentially avoided:
Roadway damages, e.g. unevenness, ruts, and pot holes 
Obstacles on the road, such as broken down vehicles 
Excessive speed in curves, especially in irregular road conditions 
Friction steps caused by oil, gravel, sand and bitumen.

Motorcycles are statically unstable and only become stabilised in motion. The stabilisation of these two wheeled vehicles is mainly achieved through steering, which is designed to turn into the lean of the vehicle. The resultant centripetal acceleration tends to correct the lean in much the same way as a controlled inverted pendulum. 
These mechanisms rely on the availability of sufficient friction between the tires and the road. If a lower frictional force is available, as for example in slippery conditions, motorcycles can become irreversibly unstable, unlike four wheeled vehicles.
Considerable progress has been achieved with the introduction of vehicle control systems, such as the Bosch Motorcycle Stability Control (MSC) and Antilock Brake Systems (ABS). However these systems control braking and accelerating, mainly by monitoring wheel speed. 
Therefore, there is scope for research in situations where loss of traction is not due to thrust or braking, but due to lateral slip.

Project Objectives
The main objectives of this project were to:

a. Provide a background about the behaviour of motorcycles and identify parameters from which lack of stability during turning could be identified.
b. Design and construct an electronic data capturing system for a test rig intended to collect the identified parameters. 
c. Install and test the electronic data capturing system under different operating conditions.

Project Methodologies

Theoretical analysis identified that the following measurements could be used to identify slip:
Vehicle Roll Angle, Angular Velocity and Lateral Acceleration
Steering Position 
Steering Torque Measurement
The practical setup was to include a twelve inch wheel bicycle. This would be inclined at a progressively increasing angle until slipping occurs, while the frame’s motion is being observed.

Results and Achievements
The system is being assembled and tested. Throughout the development work, it was shown that the system was successful in measuring the required parameters reliably. 

References
[1]. European Commission, Traffic Safety Basic Facts 2011 - Motorcycles & Mopeds, http://ec.europa.eu/transport/road_safety, Accessed February 2013.
[2]. Lattke, B., Müller, T., Winner, H.,  MOLIFE - Hazard Detection in a Cooperative Assistance Systems for Motocycles , Paper No. 11-0070, http://www-nrd.nhtsa.dot.gov/pdf/esv/esv22/22ESV-000070.pdf, Accessed November 2012.


Governor for Speed Control and Synchronization of Various Systems 

Student: Mauro Xuereb
Supervisor: Prof. Carmel Pulé
Co-Supervisor: Ing. Paul P. Debono

2013_I

Introduction
As automation in industry is taking over several manual processes, synchronisation between various stages of a process is increasingly becoming important. This can eliminate the need for manual synchronisation which can be tedious and inaccurate. Inaccuracy can result in damage to the components and machines. 

Project Objectives
In this dissertation, a research about the speed control characteristics of various machines and motors that are commonly used worldwide was carried out to determine the best way of controlling each system. Also an extensive simulation of a PID control was carried to determine the best way to implement this system for phase synchronisation control. 

Project Methodologies
First the principle of a speed governor and how it controls the speed of a machine through a closed loop system was explained in some detail. A mechanical way to control speed is by a centrifugal governor while electronically the most used system nowadays is a PID controller. A PID was implemented to phase synchronise two permanent magnet DC motors as shown in figure 1. The speed of one motor was used a reference while the PID controls the speed of the other motor. A set of simulations was done with different settings for the PID and different loads on the DC motor to analyse the system response. 

Results and Achievements

When the final PID control system was tested it was found out that the only way that two motors will be synchronised is when they are initially synchronised. The PID is only capable of correcting the phase change only when the two motor are phase synchronised before a load is connected. It was decided that a differential synchro connected to two other synchros will be used. The differential synchro will only give the phase difference between the two rotating motors and when integrated, the output is summed up with the PID controller output to correct the phase error between the motors.
From this dissertation one can understand the true meaning of the velocity, acceleration and position states when using differential equations. It is easier to understand their meaning when a position control system is taken into account. When these states are translated into a velocity control system one tend take some time to get used to meaning of the new states. While this project was meant to produce a practical understanding of the engineering requirements it was also treated as a study how to transfer knowledge when one is dealing with states in controlling speed rather than position.
It was also concluded that it is easier to control and synchronise heavy systems with high inertias since their long time constants make it easier for the controller to follow the reference input. Systems with small inertias tend to become unstable when the output oscillates about the reference input.


Calendar
 
 
Last Updated: 2 September 2013

Log In back to UoM Homepage