Filters
Availability
Publication Year
Resource Type
Faculty
School/Institute
Publication Search Results
Sort
Results per page
1 - 5 of 5
-
(2022)ThesisUnmanned aerial vehicles (UAV) usage is constantly on the increase. Future skies have a risk of being congested with busy UAVs assisting humans in many different ways. Such congestion could lead to aerial collisions. To avoid disastrous situations, potential for aerial collisions should be addressed. Avoiding aerial collisions has been reported in various different ways in the literature. Out of all the ways available in the literature, collision cones have the ability to predict a future collision beforehand with a low computational burden. Many variants of the collision cone approach have been proposed for various different collision avoidance tasks in past research. However, avoiding a collision will have an effect on the total mission time. In spite of the large volume of past work, time-efficient collision avoidance has not been examined extensively in collision cone literature. This research presents methodologies to avoid aerial collisions in a time-efficient manner using the collision cone approach. The research in this thesis has considered all possible scenarios including heading change and speed change, to avoid a collision. The heading based method was mathematically proven to be time-efficient than the other methods. Initially, 2D collision avoidance methodologies are presented; however, in extreme cases, 3D collision avoidance is necessary and 2D methods have been extended to address 3D collisions. The proposed heading based method was compared with other works presented in the literature and validated with both simulations and experiments. A Matrice 600 Pro hexacopter is used for the collision avoidance experiments.
-
(2022)ThesisWith the rapid growth of space technology, space robots play a critical role in on-orbit servicing missions, such as assembling, repairing, refueling, and transporting missions. Space robots can autonomously carry out on-orbit missions, avoiding dangerous and expensive tasks for astronauts. Unlike ground robots with fixed bases, the coupled dynamics between the free base and the manipulator of space robots need to be considered. Compared with single-arm space robots, dual-arm space robots can implement more complex tasks with a higher probability of success. Therefore, the modeling, motion control, hybrid position/force control, and post-capture control of a dual-arm space robot are investigated and presented in this thesis. The mathematical models of dual-arm space robots are developed by considering the reaction wheels (RWs) in the base. The kinematic model is constructed by the Generalized Jacobian Matrix (GJM). The dynamic models are inferred by the Newton-Euler method and the Lagrangian method, which are used in different application scenarios. The motion control of the two manipulators is used to implement a novel strategy to approach a defunct spinning target in space. By the nonlinear model predictive controller (NMPC), the end-effectors can track and plan smooth trajectories to approach and synchronize with a defunct spinning target. Meanwhile, the base attitude is regulated by the RWs to be stable at zero. The hybrid position/force control is applied to the dual-arm space robot to conduct contact operations. Novel capture and on-orbit assembly strategies are investigated. With the model uncertainties of the space robot, a robust sliding mode controller (SMC) is developed for better robust performance than the conventional computed torque controller. Furthermore, the unknown inertial parameters of the target can be precisely estimated during the capture phase. When a space robot and a target are rigidly connected during the post-capture phase, they form a combined system. The combined system can be stabilized to rest status by the space robot. The space robot can also release the target at the desired velocity. The proposed modeling, capturing of a spinning target, on-orbit assembling, and post-capturing processes are validated in the numerical simulations, which show the feasibility and effectiveness. The proposed work will improve the accuracy and efficiency of space robot technology.
-
(2022)ThesisDeep drawing is one of the most important sheet metal forming technologies for the production of numerous kinds of thin-walled metal components such as automobile body panels. However, in a deep drawing process, the high contact stresses between the die and workpiece surfaces often bring about scratching damage to the counterpart surfaces. Over the past many decades, manufacturers have found extreme difficulty in predicting the development of surface damages, such as the depth evolution of surface scratches, caused by the coupled effects of various factors in deep drawing processes. Therefore, this thesis aims to investigate the surface damage mechanisms in contact sliding and explore the application of artificial intelligence methods in the detection and prediction of surface damage. This thesis starts with an experimental investigation into the influence of tool and workpiece properties on the wear behaviour. Tribological experiments are carried out by concentrating on the differences in the friction coefficient, surface morphology and material wear. The impacts of tool materials, surface treatment processes and surface hardness are analysed, and the tribological properties of two advanced high-strength steel (AHSS) workpieces are discussed and compared. In addition, by conducting comparative experiments, this thesis also explores the role of wear debris in the surface damage evolution and reveals the influence of debris distribution and size variation on the tribological behaviours. To predict the evolution of surface scratching in sheet metals subjected to contact sliding, this thesis has successfully developed an intelligent prediction method based on fuzzy logic approach. Critical parameters are taken as the fuzzy variables to assess the contributions of individual variables to the surface damage. In addition, the quantum-behaved particle swarm optimisation algorithm is further developed by introducing adaptive control operators to refine the fuzzy prediction model. The verification results show that the optimised fuzzy model can predict the evolution of the surface scratching damage with smaller prediction error and error deviation. Afterwards, a lightweight convolutional neural network (CNN), called WearNet, is developed and trained for surface scratching detection in contact sliding. A customised convolutional block and a global average pooling layer are used in the WearNet, which enables the reduction of the network parameter number and the improvement of computational efficiency. The network response and decision mechanism are examined by using the t-distributed stochastic neighbour embedding function and the gradient-weighted class activation mapping technique, respectively. This thesis also compares the developed WearNet with other advanced CNN-based networks. The comparative tests demonstrate the advantages of WearNet in model size, computation efficiency as well as classification accuracy. Lastly, a hybrid data-driven approach is proposed to predict the remaining useful life (RUL) of forming tools. Ball-on-disc sliding contact will be used to mimic the contact conditions encountered in metal forming processes. The lightweight WearNet is retrained using transfer learning to detect the wear states of workpiece surfaces under new contact conditions, then the detection results and other relevant input parameters are incorporated into a regression model for RUL prediction, in which the bidirectional long short-term memory (BLSTM) network is employed. The prediction performance of the hybrid approach is evaluated and compared with other state-of-the-art methods to demonstrate its effectiveness and superiority in complicated RUL prediction problems.
-
(2022)ThesisBody temperature is a primary health marker of metabolic health, circadian rhythm, physiological activities, infection, and disease. Accurate and real-time monitoring of skin temperature changes offers a new way for disease diagnosis, infection monitoring, fitness tracking, and athlete performance. Wearable temperature sensors for healthcare and fitness monitoring are required to meet the high resolution (±0.1-0.2 °C) of medical-grade thermometers. However, achieving this goal has proven challenging because wearable sensors have been found to respond strongly to mechanical deformation, primarily associated with skin-stretching and touch pressure. This thesis presents the progresses that I have made towards high-accuracy wearable temperature sensors by minimizing cross-interferences of wearable temperature sensors. To achieve this aim, two complementary strategies have been developed: (1) increasing the temperature sensitivity and (2) reducing the deformation-induced cross-interferences. These strategies have been applied to two different types of sensors: resistive and capacitive. In the first strategy, several approaches have been used to enhance the sensors’ sensitivity depending on their sensing mechanisms, as briefly summarized below • New methods for controlling the density and pattern of microcracks have been developed to improve the temperature sensitivity of a resistive sensor, which consists of a Poly(3,4-ethylenedioxythiophene) Polystyrene Sulfonate (PEDOT:PSS) sensing layer deposited on a thin Poly(dimethylsiloxane) (PDMS) substrate. The temperature sensitivity is found to be strongly influenced by the crack morphology in the PEDOT:PSS layer, which can be controlled by three fabrication parameters: pre-stretching strain, substrate roughness, and acid treatment time. The maximum temperature sensitivity achieved is 0.042 °C-1 when crack length is 185.2 μm and crack density is 22.84 mm-1. This is a 42-times improvement as compared to the reference sensor without microcracks. • New methods have been developed to improve the sensitivity of capacitive temperature sensors by rational design of dielectric separates. Since the sensitivity of capacitive temperature sensors depends on the coefficient of thermal expansion and the temperature coefficient of permittivity, it is found that by selecting the low-melting-temperature (65°C) thermoplastic polyurethane (TPU) with significant thermo-responsive dielectric properties around skin temperature (30 – 45 °C) as the dielectric layer, the temperature sensitivity can achieve 0.007 °C-1. Replacing the TPU dielectric layer with Polyvinyl Alcohol (PVA)-based organogel, the temperature sensitivity can be further improved to 0.093 °C-1. In the second strategy, the following new approaches to suppress strain and pressure interferences have been investigated. • Patterning the electrodes of capacitive sensors into kirigami cut to suppress the cross-interference of stretching. The electrodes of capacitive sensors consist of silver nanowires (AgNWs). The kirigami pattern is demonstrated to effectively reduce strain interference. Computational modeling of kirigami geometry’s effects on strain and temperature sensitivities of capacitive sensors reveals that tailoring kirigami design can reduce the strain cross-sensitivity by a factor of 3125, which enables the capacitive temperature sensors to achieve the resolution of 0.14 °C. • A pressure-insensitive supercapacitive sensor has been develop by synthesising PVA organogels consisting of Na+ and Cl- ions to be used as redox-active separator. The resulting temperature sensors show very low sensitivity to surface pressure. The results reveal that the pressure interference is reduced to 0.068% (with the external pressure of 10 kPa) and the wearable temperature sensor can achieve an accuracy of 0.2 °C, matching typical medical grade thermometers. In summary, this thesis presents several major advances in creating high-accuracy wearable temperature sensors that can match the detection limits of of medical-grade thermometers. The high-precision wearable sensors have been demonstrated to offer reliable monitoring of daily skin temperature rhythm when subjected to stretch exerted by body movements and the pressure from medical compression garment. The results show that sensors can perform skin temperature monitoring with a minor error of 0.1°C compared to infrared devices. The new insights and understandings of the strategies to suppress cross-sensitivities of wearable temperature sensors to in-plane stretch and out-of-plane pressure will generate cross-cutting benefits to the design of wearable health sensors.
-
(2022)ThesisStructural composite supercapacitors have been investigated as a promising weight-saving technology for electrical vehicles (EV), electric aircraft, and mobile robots. The main objective is to maintain excellent (ideally the same as the existing structure of the same weight) mechanical properties while storing adequate electrical energy. This thesis aims to develop structural composite supercapacitors with both outstanding mechanical properties and electrical energy storage performance. The main findings and contributions of my research presented in this thesis are: (1) A novel structural electrolyte made of carbon nanofibers, epoxy, and ionic liquid (IL) that offers ionic conduction properties as well as mechanical stiffness and rigidity. The incorporation of carbon nanofibres (CNFs) into epoxy-ionic liquid-based electrolytes creates pathways for ion migration, resulting in a 40-fold boost in the ionic conductivity for the resulting electrolytes. The tensile strength and Young’s modulus of the resulting electrolytes exhibit only a slight drop. Therefore, the new solid epoxy-based electrolyte offers great potential for use in energy storage structures, for example structural composite supercapacitors and/or batteries; (2) A structural composite supercapacitor consisting of the high-performance electrodes made by grafting manganese dioxide onto carbon fibre fabrics and the epoxy-ionic liquid electrolyte. Mixing 40 wt.% of IL and 60 wt.% of epoxy (denoted as the 40IL electrolyte) yields the best combination of ionic conductivity and tensile properties. A structural composite supercapacitor has been fabricated using a 40IL electrolyte with high-capacity manganese dioxide coated carbon fibre electrodes. The resulting composite supercapacitors demonstrate excellent mechanical and electrochemical performance compared to the literature data. (3) A novel silane treatment method to enhance the ionic conduction between the electrolyte and the electrodes. The results show that the silane treatment enables the composite supercapacitors to achieve a 3-fold increase in areal capacitance without deterioration of the mechanical properties. Finally, potential opportunities for future studies of the structural composite supercapacitors are discussed.