Teaching

Project title:The fast and the furious: Artificial Intelligence-based autonomous pilots for drone racing

Project description:  As intelligent robots become more prominent in the transport and entertainment industry, researchers go to great lengths to bring human-like maneuver skills to micro unmanned aerial vehicles. Drone racing is a great application for that problem, as it is a competitive sport in which autonomous drones race against each other, while also trying to beat professional pilots. Artificial intelligence, and more specifically computer vision, might be a game changer for this sector, where traditional rule-based approaches suffer from their rigidity. In this thesis, instead of dealing with perception-planning-control problems separately, a deep learning approach is used, where RGB images are used as input for a deep convolutional neural network, which outputs reference velocities for the drone on the x-, y-, z-axes. To keep track of the drone velocity at all times, visual-inertial odometry is utilized.

Project title: Implementation and Test of Retrospective Cost Adaptive Controller for a Quadcopter with Varying Configuration Parameters

Project description:  Nowadays, UAVs are widely used for multiple applications in both, civil and military sectors. Based on each application, a specific set of dynamics is desired in order to successfully accomplish the assigned tasks. However, the design and development of specific control techniques could be tedious and time consuming and commonly not suitable for other applications. Consequently, a single flexible platform that is capable of adjusting to meet the requirements for a number of applications becomes an ideal solution for this problem. One way to achieve this goal is the implementation of adaptive controllers. The Retrospective Cost Adaptive Controller (RCAC) is a technique that makes use of a Recursive Least Square algorithm. This algorithm generates optimised inputs which are the result of the minimisation of the error defined by the cost function. In this thesis, decentralised RCACs for position control were implemented into a dynamic model of a quadrotor and its capability to reject disturbances and follow commands was analysed. For this study, constant and harmonic commands and disturbances were used for each axis and the tuning parameters which led to the best response were identified.

Project title: Collective Dynamic Response of a Leader-Follower Networked System Governed by Distributed Consensus Dynamic

Project description: In the advent of building an artificial swarm (50) of buoys – robotic floating bodies – we encountered an engineering problem. In a leader – follower architecture of swarming, the agents participating in consensus process can be driven to opposing direction by multiple leaders. This sometimes create a deadlock in robotic swarms, if, the leaders have equal and opposite input magnitude. Unable to aggregate, or explore, their mission is compromised. The solution of course is to integrate a decision making model such as Majority vote model. But, that requires the distributed sparsely connected system to have a complete knowledge of itself. This lead us formulate the problem as that of science. How does a system that participates in consensus handle conflict? How does network topology affect this process? If Network topology affects the dynamics of system, can we model the topology to minimize the effect of conflict? This study thus explores the intertwining dynamics of consensus and conflict in Multi-agent systems.

Project title: Mathematical Modelling and System Identification of Tilt-rotor Tricopter

Project description: With multicopters finding more and more application in various fields, the study of different configurations to design a better controller is an active research topic. This project aims to identify a model and estimate unknown physical parameters of a tilt-rotor tricopter UAV. Integrating an on-board computer (Raspberry Pi 3) with the UAV is done for autonomous indoor wireless control using the OptiTrack Motion Capture System for tracking the UAV. The process of system identification requires input/output data to determine the system. Identifying a model of the UAV is crucial in design of model based controllers and also estimation of the unknown physical quantities. A non-linear model of the tricopter is dervied and the significance of certain drag coefficients in the modelling of this tricopter UAV is studied. Identification and parameter estimation is done on simulation and real-time data using MATLAB and Simulink.

Project title:  Deep neural network for improving trajectory tracking of unmanned aerial vehicle

Project description: Conventional approach to control UAV is constrained by uncertainties and hidden dynamics of quadrotors. This project aims to improve trajectory tracking of UAV with deep neural network. Deep neural network learns from large dataset and discovers intricate structure by back-propagation algorithm. First, a large set of training data is collected when the flight of UAV is controlled by PID controllers. Next, deep neural network would be trained offline with the dataset. After offline training, deep neural network would learn online and get real-time update from gradient descent. Finally, the neural network structure and the learning targets should be modified if the tracking error is not in an acceptable range. With deep neural network learning structure, the UAV should be able to remain high stability and accuracy in more complex working environment.

Project title:  Quarter car modelling and control

Project description: In recent years, Autonomous Cars have been developed rapidly and even have been put on US roads without a safety driver by Waymo (the autonomous vehicle division of Alphabet, Google’s parent company) in 2017. However, there are still some problems need to be discussed and solved in order to let everyone to accept this new concept transportation. In my URECA project, I am going to scale down the autonomous car by using an RC car and simulate the scenario in real word such as overtaking, line detection, traffic sign recognition and so on. In this method, I would like to get more familiar to how a autonomous vehicle works and how to improve its performance and functionality.

Project title:  Design of Unmanned Aerial Vehicle Delivery Box

Project description: In the recent years, unmanned aerial vehicle has been developing very quickly and is able to carried out difficult task such as information gathering, search and rescue and even delivering of goods. This project will aim to develop a storage box that will have automated opening of doors and locking system to attached to the drone. The opening and locking mechanism will be design and develop using 3D printing. A servo will be attached to the 3D printed parts and controlled using Arduino.

Project title:  Development of a Multi-Rotor VTOL UAV: Control of a Multi Rotor VTOL UAV for a Variable Payload Location

Project description: To ease the traffic congestion in the foreseeable future and also to introduced a new way of transport in Singapore, a Vertical Take-off and Landing (VTOL) Unmanned Aerial Vehicle (UAV) is being suggested. Due to the lack space for a run way, the VTOL capability is the most suited for this application. It is proposed that the UAV will be modular in certain places such as the propulsion system, which will be using either hybird electric engine or fully electric motors. The cargo compartment will also be modular in a way which we could swap the module in the cargo hold between passengers module or cargo module. The UAV will have a payload of roughly 150 kg and a capability to fly across the whole length of Singapore which is roughly equal to 50km in one single flight.

Project title:  Internal Structure Design and Manufacturing of 3D-Printed Unmanned Aerial Vehicle (UAV)

Project description: This project aims to: (1) develop two different types of cellular internal structures to improve the bending stiffness and (2) reduce the mass of a vertical take-off and landing (VTOL) UAV. The bending of UAV wings can be closely modelled after the bending of cantilever beams, which are designed and 3D-printed with two different cellular structures (Diamond Honeycomb and Three-dimensional Kagome Structures). These cantilever beams are fabricated using a Stratasys Fortus 450mc Fused Deposition Modelling (FDM) machine with Acrylonitrile Butadiene Styrene (ABS). To evaluate its bending stiffness, simultaneous bending tests and numerical simulation were carried out using the Shimadzu Autograph AGX-Plus machine and the Abaqus Finite Element Analysis (FEA) Software respectively. Both the experimental and numerical simulation results provided insights on the ideal structure to use in the designing of the UAV, which was then 3D-Printed, assembled with its electronic components and tested for vertical take-off flight.

Project title: Design and Manufacturing of a multi-DOF robotic arm for rotary-wing UAV

Project description: This project aims to design and manufacture a robotic manipulator for inspection to difficult-to-access areas of civil infrastructures such as bridges. To adapt to different complex building surfaces, a 3 DOF stewart platform is designed. It is capable of performing roll, pitch and ascend/descend movements, which is important when physical contact is needed to inspect surfaces that is curved or irregular. Remaining DOFs will be provided by the UAV attached. A 3-Jaw gripper will be placed on top of the platform and both of them will be placed over a UAV. Installing gripper enables convenient change of sensors for different types of surfaces. Hence, different types of surfaces with irregular condition such as bumps and cracks can be inspected thoroughly. This is especially useful for areas such as surface under the bridge, which needs alot of manpower if direct inspection is performed.

Project title: Design and development of a gripper for unmanned aerial vehicle

Project description: UAV is currently make us more convenient. It can be used in package delivery, aerial photography. Now, we design a 3 DOF of gripper may help us to grasp something that we may not be approached such as badminton on the tree or deliver of food and first-aid equipment for the victims who stuck below canyon. The gripper on the UAV can perform 3 degree of freedom, which is ascend and descend of gripper, yaw rotation and the open-close of the claw. We use the microcontroller on the UAV to program the gripper. Hence, when we need to switch our gripper to another UAV, we only need to connect all the port to the microcontroller and download the command of gripper to the new UAV’s microcontroller. Finally, flight test will be tested to ensure the gripper functions properly as we expected.

Project title: Implementation and integration of a closed-loop system for industrial external axes on robotized paint applications

Project description: The objective of this project is to propose and integrate, using the necessary communication network open protocol, external systems and ABB controllers, capable enough to drive low weight objects. This system must be controlled through an Human Machine Interface (HMI). With a growing demand to expand automation and control across all industrial processes, communication among machines, devices and peripherals, has become a necessity. Communication allows improved collaboration among machines and reduction of dead times providing more efficient outcome under a lower cycle time.The User Interface (UI) should permit controlling the external devices at will, under operating ranges.

Project title: Fly Without Borders with Additive Manufacturing: A Microscale Tilt-Rotor Tricopter Design

Project description: During a rescue mission, multicopters offer safe surveillance, mapping and delivery of light supplies to the inside of a collapsed structure for an outside rescuer. For greater functionality, the multicopter should be agile and small enough to maneuver tight spaces, and customized to carry necessary electronics to meet operational requirements. Bearing a smaller propeller area and superior flight dynamics over other multicopters, a 3D-printed microscale tricopter offers a customizable, lightweight, and compact solution to this setting. As such, this project explores a lightweight and fully customized fabrication of a 3D-printed microscale tricopter. Considering the features and flight electronics required, the microscale tricopter frame is designed and optimized in SOLIDWORKS to compactly incorporate the electronics with minimal parts. For the desired strength and durability, the parts are printed with Acrylonitrile Butadiene Styrene (ABS) in the CUBICON 3D Printer. Tri-blade propellers are also used for a smaller blade area to keep the system compact. These methods produce a successful and stable micro-scale tri-copter weighing 519g and sizing 176mm (Radius) by 10mm (Height). Furthermore, we also demonstrate its tracking control in an indoor motion capture environment utilizing a nonlinear model predictive control (NMPC) framework, which is implemented on an on-board Raspberry Pi 3 embedded processor.

Project title: The Development of the ROS-Gazebo Package for a Rotary-Wing UAV Equipped with a Robotic Arm

Project description: Adding a robotic arm to the UAV will enhance the UAV ability to perform object manipulation, and thus expand the usage of the UAV. In this project, the motion planning algorithms of a redundant system of a UAV and an arm is developed. The redundancy of the system is utilized for additional tasks, such as obstacle avoidance and joint limit constraint. Furthermore, a camera is also attached to the system to detect the Aruco marker on the target to provide the pose of the target. All of these are implemented in the ROS-Gazebo to create a simulation package for the system.

Project title: Design of Vertical Take-off Landing Quadplane

Project description: This project aims to design and implement a tilting mechanism using additive manufacturing (3D Printing) for a fixed-wing hybrid-VTOL quadplane to allow for transition between vertical hover to forward flight and vice versa. Current fixed-wing VTOL planes employ a push-rotor mounted at the rear and switch between the vertically-mounted rotors and rear rotor for vertical and forward flight respectively. The tilt mechanism allows for a more efficient flight as it eliminates the dead weight as well as a reduction in drag from a windmilling rotor in flight.

Project title: Computational Fluid Dynamics(CFD) Analysis of a 3D-printed UAV

Project description: The project is about simulating a given UAV model and finding the aerodynamic derivatives of the design. The aerodynamic derivatives would be obtained around the suggested trim angle and speed that is given by XFLR5 and the results obtained through simulation would be compared against XFLR5 results to determine the validity of XFLR5 to obtain the aerodynamic derivatives.

Project title: Design and manufacturing of a multi-DOF lightweight aerial manipulator for rotary-wing UAV

Project description: This project aims to design an experimental setup of the robotic manipulator for UAV which is being designed in SolidWorks and later manufactured them using a 3D printer. The ability of UAV to manipulate objects may significantly expand the application area of UAVs. UAVs with the dexterous arm can lead to innovative applications such as infrastructure repair, agricultural care and possibly even construction or assembly. The UAV will be equipped with 4 degree of freedom (DOF) manipulators. The manipulator design consists of a base attached to the UAV, a robotic arm with multi-DOF, and a gripper-style end-effector specifically developed for various type of grasping.

Project title: Dual Rotor VTOL Fixed-wing UAV with 3D Printing

Project description: Vertical take-off and landing (VTOL) fixed-wing unmanned aerial vehicle (UAV) combines the advantages of fixed wing aircrafts and rotary aircrafts – the fixed wings guarantee high aerodynamic efficiency while the VTOL capability reduces reliance on runway. We have built a dual rotor VTOL fixed-wing UAV with two rotors by using additive manufacturing technology (also known as 3D Printing) as the primary manufacturing method. The high strength and rapid prototyping features of 3D printing are utilized in our design to achieve adequate structural strength without inducing weight penalties. The dynamics and control of the UAV in both VTOL mode and fixed-wing mode have been analysed and implemented to achieve ideal flying qualities. Multiple prototypes have been developed and flight tests have been conducted to verify the design and the feasibility of 3D Printing in manufacturing VTOL UAVs.

Project title: Design of process monitoring system of extrusion based 3D printers

Project description: This project aims to reduce time and material wastage by developing a process monitoring system that uses deep neural networks to identify errors autonomously that occur during a 3D print job. A print can take up to 8 or 9 hours to complete and as with any other manufacturing process, errors may occur during the print. Without human observation, the print will continue regardless of any errors happening as the printer is unable to detect them. Various forms of errors such as under extrusion, misalignment, stringing and sagging were simulated using actual print jobs and image altering software. Images of these errors were then organised into data banks to train the deep neural networks on the identification of errors. With this process monitoring system, the print can be monitored in real time by cameras installed on the printer. Concurrently, deep neural networks that have been trained will be able to identify errors that occur during the print and appropriate corrective measures can be taken to stop or rectify the print.

Project title: Control of a multi-DOF lightweight aerial manipulator for rotary-wing UAV

Project description: In this project, the control algorithm will being developed for multi-DOF robotic manipulator which will be attached to UAV base. The control system can perform tasks such as trajectory tracking with the end-effector. Stewart platform is used for modelling UAVs instead of using UAVs itself due to safety purpose. At first, proper dimension of stewart platform were being determined and designed. In order to able track certain object , vision based control system are being designed for robotic arm . Lastly, the performance of robotic arm are being studied during tracking certain objects.

Project title: Design, Build and Control of a Gripper for UAV

Project description: A compact and lightweight 3-dof gripper is designed and built using 3D printing which will be attached onto the UAV. The gripper can have rotation, elevation and gripping motions. The gripper will be simulated in C++ to analyse the performance. After that, autonomous control of the gripper is implemented using rosserial and Arduino. Finally, real time flight tests are conducted to evaluate the performance of the gripper.

Project title: Wind Tunnel Testing of a 3D Printed Flying Wing VTOL UAV

Project description: In this Final Year Project, wind tunnel tests were conducted on a half-scaled 3D printed prototype of the UAV which was manufactured out of Polylactic Acid (PLA) using Fused Deposition Modelling (FDM). The main goal of the wind tunnel experiments were to obtain the aerodynamic characteristics of the UAV from 0° to 90° angles of attack. As the VTOL UAV transitions from hover to level flight cruise, its rotors are tilted accordingly which induces varying angles of attack on the flying wing. Hence, the high angles of attack experimental data is crucial in the study of the transition flight manoeuvre of the UAV. Methods to obtain the high angles of attack data in the wind tunnel tests were developed and discussed. Moreover, several structural and aerodynamic limitations of the wind tunnel model were also discussed whilst modifications were made to the wind tunnel model to obtain more accurate experimental data. Finally, aerodynamic mathematical models were derived based on the experimental data and inputted into a Simulink model to study the transition flight manoeuvre of the UAV.

Project title: Feasibility Studies of Techniques to Strengthen 3D Printed UAV

Project description: 3D printing is a commonly used method to design and manufacture Unmanned Aerial Vehicles (UAV). It allows for UAV parts to be highly customized to suit a wide variety of applications, such as surveillance, medical aid, search and rescue. However, the size of UAVs are larger than most desktop 3D printers currently available, hence the print is usually broken out in segments and printed along the Z-axis. Such printing method would result in weakening of structural strength, which will be hazardous to objects underlying if failure occurs during flight missions. This project aims to investigate the feasibility of 3D printing techniques to strengthen 3D printed UAV parts, especially to prevent failure of wing structures under aerodynamic loading. The techniques being tested are post-processing methods, namely heat treatment and protective coating, as well as multi-material printing, with the PLA being the base material of comparison. Heat treatment is being done at above glass-transition temperature, while different types of surface finishes are used as a coating layer. Multi-material printing is being experimented using stronger/flexible filaments to improve flexural strength and toughness.

Project title: Navigation and map representation in indoor environment for an automated construction quality assessment robot system

Project description: This project aims to develop a navigation strategy for automated construction quality assessment ground robot system that inspects an unknown indoor environment. Particle filter indoor SLAM based on 2D laserscan data and inertial sensors is used to localize and build the map of the unknown indoor environment. Then, dynamic A* path planning algorithm is employed to provide feasible paths with obstacle avoidance behavior for the UGV. As the UGV have to visit all rooms in the same level of the building, frontier based exploration will be developed to generate goals to explore unknown environment.

Project title: Fabrication and Real-Time Testing of a Tilt-Rotor Tricopter

Project description: The goal of this project is to design and manufacture a lightweight unmanned tricopter that can autonomously perform a series of manoeuvres in an enclosed test range with high degree of safety and reliability. Using computer aided design and engineering simulation software, the tricopter is designed to be robust and lightweight. The tricopter will be manufactured using a combination of 3D printed ABS plastic and woven carbon fibre prepregs to guarantee a high degree of reliability during experimental flights. In order to obtain accurate data for the experiment, a motion capture system is utilised to capture precise movements of the tricopter and relay the readings to a main control computer over Wi-Fi. The obtained data is used as feedback for the Raspberry Pi computer fixed on board the tricopter, which then sends adjustment commands to the craft to achieve the desired manoeuvre/position.

Project title: Small-size UAV design and control

Project description: The FYP Small Size Uav Design and Control aims to produce an operational small size Coaxial Tricopter (Y6) at the end of the project. The motor-to-motor distance of the designed Y6 would be lesser than 25cm. Initially, the frame of the Y6 would be designed using Solidworks. Components for the multirotor would be selected concurrently to ensure that there are no dimension mismatch with the Y6 frame. The frame is to be manufactured via 3D-printing. Flight test would then be conducted with the Y6 using low level controllers. PID tuning would be done to ensure stability of the Y6 while hovering. After achieving stability in hover, basic manoeuvre involving roll, pitch and yaw would be performed with the Y6. The project would then be concluded performing trajectory flight with the Y6. This would be done by incorporating an on-board computer and markers into the Y6 to allow for communication between the user and the Y6.

Project title: Design of 3D Printed UAV Structure

Project description:  The objective of this final year project is to build Light weight, low-medium range fixed wing Surveillance UAV with Vertical Take-off and Landing (VTOL) configuration using 3D printing technology. With the use of the engineering software such as SolidWork and XLFR5 the design was well weighted to its designed structure. The main structure of the UAV is manufactured using 3D printed ABS plastic for wings, fuselage, and carbon fiber rods for the tail section. Combination of this both material reinforces the structural stability of the UAV. A 5-motor configuration is being used on this UAV, where 4-brushless motors for Vertical Take-off and Landing (VTOL) and a single brushless pull motor for the forward flight. The UAV operates with a higher capacity battery to increase the endurance of the flight time. Overall, with the fixed wing design and high endurance flight time the UAV will be able to cover a large area of land for surveillance purpose.

Project title: Design and fabrication of a landing wheel mechanism for a quadcopter

Project description:  The goal of this project is to design and manufacture a landing wheel mechanism for a quadcopter, enabling it to perform basic ground maneuvers without compromising it’s fight capabilities. In order to achieve this, the landing wheel mechanism was designed to be lightweight while still being able to withstand the forces during take-off and landing. The design was done through the use of computer aided design where a detailed design could be done. Thereafter, the design will be built before conducting ground and flight tests to ensure the viability of the design.

Project title: Design and Manufacturing of a VTOL UAV by 3D Printing

Project description:  The FYP aims to enhance material produced using 3D printing method to effectively manufacture UAVs with high mechanical strength. In this project, we look into strengthening methods such as fill compositing and CFRP reinforcing. In fill compositing, stronger materials are added to the pre-designed empty space of the 3D printed parts to increase their strength. As for the CFRP reinforcing, carbon fiber pre-pregs are added to the surface of the parts by the mean of adhesive to externally strengthen them. By studying those method, we hope to make 3D printed parts lighter as well as stronger and thus become more viable for usage in making UAVs.

Project title: Design, manufacturing and testing of a flying car

Project description:  This project aims at designing an unmanned aerial vehicle with ground movement capabilities. For this purpose, quadcopter frames will be integrated with the chassis of a ground vehicle without compromising their flight and ground capabilities respectively. The design has to be lightweight and concise to enable efficient take-off and landing with sufficient force and minimal power consumption. Subsequently, prototypes will be manufactured and flight and ground tests will be carried out to obtain the most efficient design.

Project title: Pan-Tilt Camera control for precise landing of UAVs

Project description:  I will work on Part 2 of this project, which is to design a platform structure for the pan-tilt camera. The main objective is to allow the pan-tilt camera to maintain an accurate movement and as light as possible. The process of Part 2 consists of 4 main parts, which are: 1. Platform structure design, 2. Building prototypes using Additive Modelling (known as 3D printing), 3. Installing servos (or brushless motor), camera for testing and 4. Finalise the design by optimize the performance of the platform designed. The first prototype (printed with PLA, uses analogue servos) has been built and tested, it is functioning but needs a higher accuracy and more precise motion to fulfil the goal of this project. A better model will be built using a new design and installed with digital servos. If it is necessary, different materials and manufacturing techniques (other than Fusion Deposition Modelling, FDM) might be utilised for a better result.

Project title: Pan-Tilt Camera Stabilization and Control

Project description:  This project aims to control the pan-tilt camera, which includes the stabilization and vision-based control (VBC) modules. In the stabilization module, inertial information from IMU will be applied for stabilizing the camera using gimbal. In the VBC module, the camera will always point at a pre-defined or un-defined target object in the precise landing of UAV by using visual tracking algorithms.

Project title: Design and Manufacturing of a Tilt-Rotor Bicopter Experimental Setup

Project description:  The initial phase of this project is to first design an experimental setup needed in order to house and securely mount a bicopter which is to be used in later stages to run tests to determine the flight controls. The design of the experimental setup (cage) should allow for the yaw, roll and pitch features of the bicopter. 2 design perimeters are implemented into the design of the experimental bicopter that involves both the oblique tilting feature (maximum tilting angle of 30 degrees) and also a tilting feature of the motors along the lateral axis of the bicopter. CAD drawings are initially drafted to produce a virtual design of the entire setup. Multiple manufacturing processes such as 3D printing and several CNC machining steps are involved in the production of the setup. Upon successful production of the setup and bicopter, flight control inputs are inputted into the PixHawk to perform test to determine the flight responses from the experiments. Such evaluation of the flight responses are determined and obtained via torque measurements. Hence the main motivation behind this project is first to utilize any available manufacturing processes to produce a tangible experimental setup and also to study on how the effect of the oblique tilting features influences the flight responses of the bicopter.

Project title: Sensory system design for an Automated Construction Quality Assessment Robot System

Project description:  The issues of the quality of a house is paramount for the owner in Singapore. The quality of the house is subjected to defects such as hollowness, crack, unevenness etc. To resolve these issues, the BCA is using a comprehensive method called CONQUAS to evaluate the quality of the house. Despite the increasing interest in academic, having a robotic system for CONQUAS are still very scarce. Therefore, this report intended to design and construct a robotic system to aid in the inspection. This robotic system will consist of a set of imaging camera and sensors to evaluate the quality. This design, which was similar to a trolley architecture, allow integration and interoperability between the mobile robots and sensor. It is the first time that such a robot is designed for inspection. Furthermore, it is able to become standalone when mobile robot is malfunction, hence making this system more durable. Likewise, this report will mainly focus on the process of the designing the platform and the motion of the motor control.

Project title: Design and realization of a twin rotor system

Project description:  This final year project presents a Twin Rotor System model with the application of PIXHAWK and MATLAB Simulink®. Mechanical design on Twin Rotor System experiment setup will be reviewed based on several Basic Design Rules and Embodiment Design Principles. Besides, components involved in the architecture of circuitry for Twin Rotor System experiment setup will also be described in detail. The purpose of designing Twin Rotor System is to establish a data acquisition experiment setup. In this 1 degree of freedom Twin Rotor System experiment setup, control on Roll – angle are interested. System model of Twin Rotor System will be constructed by means of MATLAB Simulink®. PID tuning of Twin Rotor System which gives the most pleasing response are recorded and the appropriateness of Twin Rotor System response will then be evaluated by plotting graphs of angular displacement over time.

Project title: Precise landing for unmanned aerial vehicles for an undefined target object

Project description:  Firstly, the OpenCV library used for these techniques is described. Then, the concepts of detectors and descriptors are introduced. Detectors find points of interest in an image, and descriptors are used to describe what an image patch around these points are like in a compact form. Following, there are then details on how these algorithms are implemented in OpenCV, and how the Python interface works. The hardware used and the steps to install and set up the software are then detailed.

Project title: Precise landing for unmanned aerial vehicles for an pre-defined target object

Project description:  The goal of this thesis is the autonomous landing of an unmanned aerial vehicle(UAV) on a predefined area. A vision-based approach was chosen for landing target recolonization, with a single on-board camera being used. Identification of the landing target, and pose estimation of the UAV’s relative pose to it, are all done through computer vision. An algorithm was developed for estimate the orientation as well as relative position of the landing target to the UAV using visual markers on the landing target as well as computer vision. In the end, tests are performed using both simulations as well as a UAV, in which autonomous landing is achieved. This paper also compares and tests different fiducial marker seizes, to seek the most appropriate to use in order to accomplish the task of autonomous UAV landing. The tests seek to identify differences in performance of different size markers under variables such as distance from camera and degree of rotation. All tests on fiducial marker detection are done using the popular ArUco framework.

Project title: A Simulation Study on Fuzzy Logic Control of a Quadcopter UAV using a PSO-SMC Hybrid Learning Algorithm

Project description:  In this final year project, a model-free Takagi-Sugeno-Kang Type-2 fuzzy neural network (T2-FNN) is developed and trained using a hybrid learning algorithm based on particle swarm optimization and sliding mode control. The proposed T2-FNN is implemented on a typical quadcopter in a virtual environment and its trajectory tracking performance is evaluated. In order to emulate real-world operational environments during flight, the effect of global positioning system noise on the vehicle’s trajectory tracking performance is also considered in addition to the ideal case without noise. Finally, a comparative study against conventional model-based proportional-derivative control is presented in order to assess the relative benefit of the proposed T2-FNN for UAV applications. Key findings from this study conclude that the proposed T2-FNN is a preferable choice over conventional control methods as it offers a non-linear, model-free approach with excellent noise-handling capability. Looking forward, future works aim to extend the use of this control technique into specific real-world applications such as surveillance and agriculture in order to fill the emerging role of UAVs in the world today.

Project title: Adaptive Neuro-Fuzzy Logic Learning-Based Control of a Quad-Rotor

Project description:  The project is intended to explored the issue of reducing UAV’s vulnerability to transient environment. During which, a model-based drone controller is designed with learning capability, which integrated artificial neural network and fuzzy logic control algorithm in simulation environment, arising as a possible new form of real time flight control. The capability of tracking real-time trajectory is verified with tolerable error and satisfied improvement is observed while comparing the performance with its traditional counterpart.

Project title: Modeling and Identification of Tricopter Unmanned Aerial Vehicle

Project description:  The initial phase of this project is to build a tricopter and provide concepts, dynamics and control strategies of the UAV. After building the physical drone, mathematical modeling and analysis of the tricopter is done with the aid of parametric equations of motion and identification. The secondary objective describes the modeling and simulation of the coaxial tri-rotor UAV using Matlab/Simulink. Simulation results are presented and the effects of controller parameter variation and external disturbance components are discussed.

Project title: Design and realization of a tilting mechanism

Project description:  In terms of cost effectiveness and power efficiency, tricopter is better than other Vertical Take-Off and Landing (VTOL) vehicles. However, tricopter control and stability is more challenging compared to even numbers configuration such as quadcopter. Since it has odd number of propellers, tricopter utilizes tilting mechanism for yaw rotation. Main focus of this research will be about tilting mechanism and yaw rotation. All the systems tested on this experiment is mounted to the ground to avoid injuries for the researchers or to the UAV. To achieve this, Twin Rotor Multi Input-Multi Output (TRMS) system is developed for early control testing. Mounting for general UAVs is also developed to test control design of any UAV. UAV mounting constraints translational movement of a UAV and reduce the desgree-of-freedom into 3 for rotational movement.

Project title: Automatic UAV Launcher With Bluetooth Low Energy Integrated Electromagnetic
Releasing System

Project description:  This final year project presents a novel state of the art catapult launcher system with Bluetooth low energy (BLE) integrated feedback system which is designed for small size unmanned aerial vehicles (UAVs). Whereas the launcher uses spring energy as the primary energy source, its carriage uses electromagnetic energy to hold and release the UAV. Project divides into two phases for the convenience of fabrication. First phase focuses on the railing system along with the main frame design. Phase two dedicates on appropriate car design with a launching mechanism and a feedback system for sensor data by using BLE. In phase one, few of the major requirements are the weight of the main frame and the strength. Therefore Aluminium is used for main frame fabrication, where stainless steel is used in areas where high concentrated stresses are found. A patent research is carried out in phase one and elaborates in later chapters. In phase two, a magnetic launching mechanism is designed along with an ultrasound sensor feedback control system. Pressure sensors are used to measure the force acting on the carriage.

Project title: Development and Implementation of Vision-based Techniques for Target Identification

Project description:  Object detection is one of the complex problems in area of computer vision. As new developments are coming through, application of object detection and localisation extends to multiple fields. One of its applications would be aerial surveillance by processing real time images to detect and locate a specific object. In this project, an aerial boat detection system is imitated and an algorithm for a simple boat detection system is presented. In this system, SURF (Speeded Up Robust Features) detector, bag of keypoints and support vector machine (SVM) algorithms are utilized. Development of boat detection system is made up of training phase and detection phase. Experimental evaluations show that system can detect and locate a floating paper boat in an image with an average accuracy of 90.1%. The system has been optimized by varying SVM’s parameters and the results are presented in this report.

Project title: Development of New Control Algorithms for High Performance Autopilots

Project description:  The project simulates dynamics of an aircraft subjected to an external environment that has a direct impact on the stability and behaviour of the aircraft. The Eurofighter Typhoon 2000 was chosen as the reference aircraft for the course of the project. The aircraft was subjected to open loop testing using MATLAB/Simulink on the nonlinear system of equations formulated by using the general six degree of freedom set of equations for an aircraft moving through space. Stability derivatives for these equations were obtained from the TORNADO Toolbox using the general geometry of the wing and simulating it as a flat plate.

Project title: Energy harvesting via piezoceramics in unmanned aerial vehicles

Project description: The process of wasted vibration is harvested and converted to useful electrical energy to power wireless sensor networks. One such effective application would be for aircraft health monitoring systems within the flying environment. The value proposition of such method is the ease and economical deployment of monitoring and measuring points in remote areas that would not be economically viable to access.

Project title: Spherical rolling robot for planetary surface exploration

Project description: Given agility, spherical rolling robots (SRRs) can access into more kinds of human-unfriendly environment. Besides that, internal hardware is well protected from external physical disturbances by the spherical shell. This makes SRR very durable. These have motivated many to focus on developing spherical bodied robots including the author of this report. This report introduced the design and development of an SRR which uses a 1-DOF internal pendulum mechanism to roll forward and backward, under closed loop control system. Dynamic analyses of the system and real time simulation results have also been demonstrated in this project.