Matching Items (90)
Filtering by
- Genre: Academic theses
- Creators: Yu, Hongbin
Description
Object detection is an interesting computer vision area that is concerned with the detection of object instances belonging to specific classes of interest as well as the localization of these instances in images and/or videos. Object detection serves as a vital module in many computer vision based applications. This work focuses on the development of object detection methods that exhibit increased robustness to varying illuminations and image quality. In this work, two methods for robust object detection are presented.
In the context of varying illumination, this work focuses on robust generic obstacle detection and collision warning in Advanced Driver Assistance Systems (ADAS) under varying illumination conditions. The highlight of the first method is the ability to detect all obstacles without prior knowledge and detect partially occluded obstacles including the obstacles that have not completely appeared in the frame (truncated obstacles). It is first shown that the angular distortion in the Inverse Perspective Mapping (IPM) domain belonging to obstacle edges varies as a function of their corresponding 2D location in the camera plane. This information is used to generate object proposals. A novel proposal assessment method based on fusing statistical properties from both the IPM image and the camera image to perform robust outlier elimination and false positive reduction is also proposed.
In the context of image quality, this work focuses on robust multiple-class object detection using deep neural networks for images with varying quality. The use of Generative Adversarial Networks (GANs) is proposed in a novel generative framework to generate features that provide robustness for object detection on reduced quality images. The proposed GAN-based Detection of Objects (GAN-DO) framework is not restricted to any particular architecture and can be generalized to several deep neural network (DNN) based architectures. The resulting deep neural network maintains the exact architecture as the selected baseline model without adding to the model parameter complexity or inference speed. Performance results provided using GAN-DO on object detection datasets establish an improved robustness to varying image quality and a higher object detection and classification accuracy compared to the existing approaches.
In the context of varying illumination, this work focuses on robust generic obstacle detection and collision warning in Advanced Driver Assistance Systems (ADAS) under varying illumination conditions. The highlight of the first method is the ability to detect all obstacles without prior knowledge and detect partially occluded obstacles including the obstacles that have not completely appeared in the frame (truncated obstacles). It is first shown that the angular distortion in the Inverse Perspective Mapping (IPM) domain belonging to obstacle edges varies as a function of their corresponding 2D location in the camera plane. This information is used to generate object proposals. A novel proposal assessment method based on fusing statistical properties from both the IPM image and the camera image to perform robust outlier elimination and false positive reduction is also proposed.
In the context of image quality, this work focuses on robust multiple-class object detection using deep neural networks for images with varying quality. The use of Generative Adversarial Networks (GANs) is proposed in a novel generative framework to generate features that provide robustness for object detection on reduced quality images. The proposed GAN-based Detection of Objects (GAN-DO) framework is not restricted to any particular architecture and can be generalized to several deep neural network (DNN) based architectures. The resulting deep neural network maintains the exact architecture as the selected baseline model without adding to the model parameter complexity or inference speed. Performance results provided using GAN-DO on object detection datasets establish an improved robustness to varying image quality and a higher object detection and classification accuracy compared to the existing approaches.
ContributorsPrakash, Charan Dudda (Author) / Karam, Lina (Thesis advisor) / Abousleman, Glen (Committee member) / Jayasuriya, Suren (Committee member) / Yu, Hongbin (Committee member) / Arizona State University (Publisher)
Created2020
Description
The realization of Silicon based photonic devices will enable much faster data transmission than is possible today using the current electronics based devices. Group IV alloys germanium tin (GeSn) and silicon germanium tin (SiGeSn) have the potential to form an direct bandgap material and thus, they are promising candidates to develop a Si compatible light source and advance the field of silicon photonics. However, the growth of the alloys is challenging as it requires low temperature growth and proper strain management in the films during growth to prevent tin segregation. In order to satisfy these criteria, various research groups have developed novel chemical vapor deposition (CVD) reactors to deposit the films. While these reactors have been highly successful in depositing high crystal quality high Sn concentration films, they are generally expensive set-ups which utilize several turbomolecular/cryogenic pumps and/or load-lock systems. An more economical process than the state-of-the art to grow group IV materials will be highly valuable. Thus, the work presented in this dissertation was focused on deposition of group IV semiconductor thin films using simplified plasma enhanced CVD (PECVD) reactors.
Two different in-house assembled PECVD reactor systems, namely Reactor No. 1 and 2, were utilized to deposit Ge, GeSn and SiGeSn thin films. PECVD technique was used as plasma assistance allows for potentially depositing the films at growth temperatures lower than those of conventional CVD. Germane (GeH4) and Digermane (Ge2H6) were used as the Ge precursor while Disilane (Si2H6) and tin chloride (SnCl4) were used as the precursors for Si and Sn respectively. The growth conditions such as growth temperature, precursor flow rates, precursor partial pressures, and chamber pressure were varied in a wide range to optimize the growth conditions for the films. Polycrystalline Ge films and SiGeSn films with an Sn content upto 8% were deposited using Reactor No. 1 and 2. Development of epitaxial Ge buffers and GeSn films was accomplished using a modified Reactor No. 2 at temperatures <400oC without the aid of ultra-high vacuum conditions or a high temperature substrate pre-deposition bake thereby leading to a low economic and thermal budget for the deposition process.
Two different in-house assembled PECVD reactor systems, namely Reactor No. 1 and 2, were utilized to deposit Ge, GeSn and SiGeSn thin films. PECVD technique was used as plasma assistance allows for potentially depositing the films at growth temperatures lower than those of conventional CVD. Germane (GeH4) and Digermane (Ge2H6) were used as the Ge precursor while Disilane (Si2H6) and tin chloride (SnCl4) were used as the precursors for Si and Sn respectively. The growth conditions such as growth temperature, precursor flow rates, precursor partial pressures, and chamber pressure were varied in a wide range to optimize the growth conditions for the films. Polycrystalline Ge films and SiGeSn films with an Sn content upto 8% were deposited using Reactor No. 1 and 2. Development of epitaxial Ge buffers and GeSn films was accomplished using a modified Reactor No. 2 at temperatures <400oC without the aid of ultra-high vacuum conditions or a high temperature substrate pre-deposition bake thereby leading to a low economic and thermal budget for the deposition process.
ContributorsVanjaria, Jignesh (Author) / Yu, Hongbin (Thesis advisor) / Arjunan, Arul C (Committee member) / Alford, Terry (Committee member) / Arizona State University (Publisher)
Created2020
Description
The study of soft magnetic materials has been growing in popularity in recent years. Driving this interest are new applications for traditional electrical power-management components, such as inductors and transformers, which must be scaled down to the micro and nano scale while the frequencies of operation have been scaling up to the gigahertz range and beyond. The exceptional magnetic properties of the materials make them highly effective in these small-component applications, but the ability of these materials to provide highly-effective shielding has not been so thoroughly considered. Most shielding is done with traditional metals, such as aluminum, because of the relatively low cost of the material and high workability in shaping the material to meet size and dimensional requirements.
This research project focuses on analyzing the variance in shielding effectiveness and electromagnetic field effects of a thin film of Cobalt Zirconium Tantalum Boron (CZTB) in the band of frequencies most likely to require innovative solutions to long-standing problems of noise and interference. The measurements include Near H-Field attenuation and field effects, Far Field shielding, and Backscatter. Minor variances in the thickness and layering of sputter deposition can have significant changes electromagnetic signature of devices which radiate energy through the material.
The material properties presented in this research are H-Field attenuation, H-Field Flux Orientation, Far-Field Approximation, E Field Vector Directivity, H Field Vector Directivity, and Backscatter Magnitude. The results are presented, analyzed and explained using characterization techniques. Future work includes the effect of sputter deposition orientation, application to devices, and applicability in mitigating specific noise signals beyond the 5G band.
This research project focuses on analyzing the variance in shielding effectiveness and electromagnetic field effects of a thin film of Cobalt Zirconium Tantalum Boron (CZTB) in the band of frequencies most likely to require innovative solutions to long-standing problems of noise and interference. The measurements include Near H-Field attenuation and field effects, Far Field shielding, and Backscatter. Minor variances in the thickness and layering of sputter deposition can have significant changes electromagnetic signature of devices which radiate energy through the material.
The material properties presented in this research are H-Field attenuation, H-Field Flux Orientation, Far-Field Approximation, E Field Vector Directivity, H Field Vector Directivity, and Backscatter Magnitude. The results are presented, analyzed and explained using characterization techniques. Future work includes the effect of sputter deposition orientation, application to devices, and applicability in mitigating specific noise signals beyond the 5G band.
ContributorsMiller, Phillip Carl (Author) / Yu, Hongbin (Thesis advisor) / Aberle, James T., 1961- (Committee member) / Blain Christen, Jennifer (Committee member) / Arizona State University (Publisher)
Created2019
Description
This dissertation aims to study the electron and spin transport, scattering in two dimensional pseudospin-1 lattice systems, hybrid systems of topological insulator and magnetic insulators, and molecule chain systems. For pseudospin-1 systems, the energy band consists of a pair of Dirac cones and a flat band through the connecting point of the cones. First, contrary to the conditional wisdom that flatband can localize electrons, I find that in a non-equilibrium situation where a constant electric field is suddenly switched on, the flat band can enhance the resulting current in both the linear and nonlinear response regimes compared to spin-1/2 system. Second, in the setup of massive pseudospin-1 electron scattering over a gate potential scatterer, I discover the large resonant skew scattering called super skew scattering, which does not arise in the corresponding spin-1/2 system and massless pseudospin-1 system. Third, by applying an appropriate gate voltage to generate a cavity in an alpha-T3 lattice, I find the exponential decay of the quasiparticles from a chaotic cavity, with a one-to-one correspondence between the exponential decay rate and the Berry phase for the entire family of alpha-T3 materials. Based on the hybrid system of a ferromagnetic insulator on top of a topological insulator, I first investigate the magnetization dynamics of a pair of ferromagnetic insulators deposited on the surface of a topological insulator. The spin polarized current on the surface of topological insulator can affect the magnetization of the two ferromagnetic insulators through proximity effect, which in turn modulates the electron transport, giving rise to the robust phase locking between the two magnetization dynamics. Second, by putting a skyrmion structure on top of a topological insulator, I find robust electron skew scattering against skyrmion structure even with deformation, due to the emergence of resonant modes. The chirality of molecule can lead to spin polarized transport due to the spin orbit interaction. I investigate spin transport through a chiral polyacetylene molecule and uncover the emergence of spin Fano resonances as a manifestation of the chiral induced spin selectivity effect.
ContributorsWang, Chengzhen (Author) / Lai, Ying-Cheng (Thesis advisor) / Yu, Hongbin (Committee member) / Wang, Chao (Committee member) / Zhao, Yuji (Committee member) / Arizona State University (Publisher)
Created2021
Description
Wide bandgap semiconductors, also known as WBG semiconductors are materials which have larger bandgaps than conventional semiconductors such as Si or GaAs. They permit devices to operate at much higher voltages, frequencies and temperatures. They are the key material used to make LEDs, lasers, radio frequency applications, military applications, and power electronics. Their intrinsic qualities make them promising for next-generation devices for general semiconductor use. Their ability to handle higher power density is particularly attractive for attempts to sustain Moore's law, as conventional technologies appear to be reaching a bottleneck. Apart from WBG materials, ultra-wide bandgap (UWBG) materials, such as Ga2O3, AlN, diamond, or BN, are also attractive since they have even more extreme properties. Although this field is relatively new, which still remains a lot of effort to study and investigate, people can still expect that these materials could be the main characters for more advanced applications in the near future.
In the dissertation, three topics with power devices made by WBG or UWBG semiconductors were introduced. In chapter 1, a generally background knowledge introduction is given. This helps the reader to learn current research focuses. In chapter 2, a comprehensive study of temperature-dependent characteristics of Ga2O3 SBDs with highly-doped substrate is demonstrated. A modified thermionic emission model over an inhomogeneous barrier with a voltage-dependent barrier height is investigated. Besides, the mechanism of surface leakage current is also discussed. These results are beneficial for future developments of low-loss β-Ga2O3 electronics and optoelectronics. In chapter 3, vertical GaN Schottky barrier diodes (SBDs) with floating metal rings (FMRs) as edge termination structures on bulk GaN substrates was introduced. This work represents a useful reference for the FMR termination design for GaN power devices. In chapter 4, AlGaN/GaN metal-insulator-semiconductor high electron mobility transistors (MISHEMTs) fabricated on Si substrates with a 10 nm boron nitride (BN) layer as gate dielectric was demonstrated. The material characterization was investigated by X-ray photoelectric spectroscopy (XPS) and UV photoelectron spectroscopy (UPS). And the gate leakage current mechanisms were also investigated by temperature-dependent current-voltage measurements.
Although still in its infancy, past and projected future progress of electronic designs will ultimately achieve this very goal that WBG and UWBG semiconductors will be indispensable for today and future’s science, technologies and society.
ContributorsYang, Tsung-Han (Author) / Zhao, Yuji (Thesis advisor) / Vasileska, Dragica (Committee member) / Yu, Hongbin (Committee member) / Nemanich, Robert (Committee member) / Arizona State University (Publisher)
Created2021
Description
Wurtzite (B, Ga, Al) N semiconductors, especially (Ga, Al) N material systems, demonstrate immense promises to boost the economic growth in the semiconductor industry that is approaching the end of Moore’s law. At the material level, their high electric field strength, high saturation velocity, and unique heterojunction polarization charge have enabled tremendous potentials for high power, high frequency, and photonic applications. With the availability of large-area bulk GaN substrates and high-quality epilayer on foreign substrates, the power conversion applications of GaN are now at the cusp of commercialization.Despite these encouraging advances, there remain two critical hurdles in GaN-based technology: selective area doping and hole-based p-channel devices. Current selective area doping methods are still immature and lead to low-quality lateral p-n junctions, which prevent the realization of advanced power transistors and rectifiers. The missing of hole-based p-channel devices hinders the development of GaN complementary integrated circuits.
This thesis comprehensively studied these challenges. The first part (chapter 2) researched the selective area doping by etch-then-regrow. A GaN-based vertical-channel junction field-effect transistors (VC-JFETs) was experimentally demonstrated by blanket regrowth and self-planarization. The devices’ electrical performances were characterized to understand the regrowth quality. The non-ideal factors during p-GaN regrowth were also discussed. The second part (chapter 3-5) systematically studied the application of the hydrogen plasma treatment process to change the p-GaN properties selectively. A novel GaN-based metal-insulator-semiconductor junction was demonstrated. Then a novel edge termination design with avalanche breakdown capability achieved in GaN power rectifiers is proposed. The last part (Chapter 6) demonstrated a GaN-based p-channel heterojunction field-effect transistor, with record low leakage, subthreshold swing, and a record high on/off ratio. In the end, some outlook and future work have also been proposed.
Although in infancy, the demonstrated etch-then-regrow and the hydrogen plasma treatment methods have the potential to ultimately solve the challenges in GaN and benefit the development of the wide-ultra-wide bandgap industry, technology, and society.
ContributorsYang, Chen (Author) / Zhao, Yuji (Thesis advisor) / Goodnick, Stephen (Committee member) / Yu, Hongbin (Committee member) / Vasileska, Dragica (Committee member) / Arizona State University (Publisher)
Created2021
Description
A new class of electronic materials from food and foodstuff was developed to form a “toolkit” for edible electronics along with inorganic materials. Electrical components like resistors, capacitors and inductors were fabricated with such materials and tested. Applicable devices such as filters, microphones and pH sensors were built with edible materials. Among the applications, a wireless edible pH sensor was optimized in terms of form factor, fabrication process and cost. This dissertation discusses the material sciences of food industry, design and fabrication of electronics and biomedical engineering by demonstrating edible electronic materials, components and devices such as filters, microphones and pH sensors. pH sensors are optimized for two different generations of design and fabrication.
ContributorsYang, Haokai (Author) / Jiang, Hanqing (Thesis advisor) / Yu, Hongbin (Thesis advisor) / Yao, Yu (Committee member) / Nian, Qiong (Committee member) / Zhuang, Houlong (Committee member) / Arizona State University (Publisher)
Created2021
Description
Machine learning (ML) and deep learning (DL) has become an intrinsic part of multiple fields. The ability to solve complex problems makes machine learning a panacea. In the last few years, there has been an explosion of data generation, which has greatly improvised machine learning models. But this comes with a cost of high computation, which invariably increases power usage and cost of the hardware. In this thesis we explore applications of ML techniques, applied to two completely different fields - arts, media and theater and urban climate research using low-cost and low-powered edge devices. The multi-modal chatbot uses different machine learning techniques: natural language processing (NLP) and computer vision (CV) to understand inputs of the user and accordingly perform in the play and interact with the audience. This system is also equipped with other interactive hardware setups like movable LED systems, together they provide an experiential theatrical play tailored to each user. I will discuss how I used edge devices to achieve this AI system which has created a new genre in theatrical play. I will then discuss MaRTiny, which is an AI-based bio-meteorological system that calculates mean radiant temperature (MRT), which is an important parameter for urban climate research. It is also equipped with a vision system that performs different machine learning tasks like pedestrian and shade detection. The entire system costs around $200 which can potentially replace the existing setup worth $20,000. I will further discuss how I overcame the inaccuracies in MRT value caused by the system, using machine learning methods. These projects although belonging to two very different fields, are implemented using edge devices and use similar ML techniques. In this thesis I will detail out different techniques that are shared between these two projects and how they can be used in several other applications using edge devices.
ContributorsKulkarni, Karthik Kashinath (Author) / Jayasuriya, Suren (Thesis advisor) / Middel, Ariane (Thesis advisor) / Yu, Hongbin (Committee member) / Arizona State University (Publisher)
Created2021
Description
In this dissertation, I described my research on the growth and characterization of various nanostructures, such as nanowires, nanobelts and nanosheets, of different semiconductors in a Chemical Vapor Deposition (CVD) system.
In the first part of my research, I selected chalcogenides (such as CdS and CdSe) for a comprehensive study in growing two-segment axial nanowires and radial nanobelts/sheets using the ternary CdSxSe1-x alloys. I demonstrated simultaneous red (from CdSe-rich) and green (from CdS-rich) light emission from a single monolithic heterostructure with a maximum wavelength separation of 160 nm. I also demonstrated the first simultaneous two-color lasing from a single nanosheet heterostructure with a wavelength separation of 91 nm under sufficiently strong pumping power.
In the second part, I considered several combinations of source materials with different growth methods in order to extend the spectral coverage of previously demonstrated structures towards shorter wavelengths to achieve full-color emissions. I achieved this with the growth of multisegment heterostructure nanosheets (MSHNs), using ZnS and CdSe chalcogenides, via our novel growth method. By utilizing this method, I demonstrated the first growth of ZnCdSSe MSHNs with an overall lattice mismatch of 6.6%, emitting red, green and blue light simultaneously, in a single furnace run using a simple CVD system. The key to this growth method is the dual ion exchange process which converts nanosheets rich in CdSe to nanosheets rich in ZnS, demonstrated for the first time in this work. Tri-chromatic white light emission with different correlated color temperature values was achieved under different growth conditions. We demonstrated multicolor (191 nm total wavelength separation) laser from a single monolithic semiconductor nanostructure for the first time. Due to the difficulties associated with growing semiconductor materials of differing composition on a given substrate using traditional planar epitaxial technology, our nanostructures and growth method are very promising for various device applications, including but not limited to: illumination, multicolor displays, photodetectors, spectrometers and monolithic multicolor lasers.
In the first part of my research, I selected chalcogenides (such as CdS and CdSe) for a comprehensive study in growing two-segment axial nanowires and radial nanobelts/sheets using the ternary CdSxSe1-x alloys. I demonstrated simultaneous red (from CdSe-rich) and green (from CdS-rich) light emission from a single monolithic heterostructure with a maximum wavelength separation of 160 nm. I also demonstrated the first simultaneous two-color lasing from a single nanosheet heterostructure with a wavelength separation of 91 nm under sufficiently strong pumping power.
In the second part, I considered several combinations of source materials with different growth methods in order to extend the spectral coverage of previously demonstrated structures towards shorter wavelengths to achieve full-color emissions. I achieved this with the growth of multisegment heterostructure nanosheets (MSHNs), using ZnS and CdSe chalcogenides, via our novel growth method. By utilizing this method, I demonstrated the first growth of ZnCdSSe MSHNs with an overall lattice mismatch of 6.6%, emitting red, green and blue light simultaneously, in a single furnace run using a simple CVD system. The key to this growth method is the dual ion exchange process which converts nanosheets rich in CdSe to nanosheets rich in ZnS, demonstrated for the first time in this work. Tri-chromatic white light emission with different correlated color temperature values was achieved under different growth conditions. We demonstrated multicolor (191 nm total wavelength separation) laser from a single monolithic semiconductor nanostructure for the first time. Due to the difficulties associated with growing semiconductor materials of differing composition on a given substrate using traditional planar epitaxial technology, our nanostructures and growth method are very promising for various device applications, including but not limited to: illumination, multicolor displays, photodetectors, spectrometers and monolithic multicolor lasers.
ContributorsTurkdogan, Sunay (Author) / Ning, Cun Zheng (Thesis advisor) / Palais, Joseph C. (Committee member) / Yu, Hongbin (Committee member) / Mardinly, A. John (Committee member) / Arizona State University (Publisher)
Created2015
Description
The integration of Distributed Energy Resources (DER), including wind energy and photovoltaic (PV) panels, into power systems, increases the potential for events that could lead to outages and cascading failures. This risk is heightened by the limited dynamic information in energy grid datasets, primarily due to sparse Phasor Measurement Units (PMUs) placement. This data quality issue underscores the need for effective methodologies to manage these challenges. One significant challenge is the data gaps in low-resolution (LR) data from RTU and smart meters, hindering robust machine learning (ML) applications. To address this, a systematic approach involves preparing data effectively and designing efficient event detection methods, utilizing both intrinsic physics and extrinsic correlations from power systems. The process begins by interpolating LR data using high-resolution (HR) data, aiming to create virtual PMUs for improved grid management. Current interpolation methods often overlook extrinsic spatial-temporal correlations and intrinsic governing equations like Ordinary Differential Equations (ODEs) or Differential Algebraic Equations (DAEs). Physics-Informed Neural Networks (PINNs) are used for this purpose, though they face challenges with limited LR samples. The solution involves exploring the embedding space governed by ODEs/DAEs, generating extrinsic correlations for initial LR data imputation, and enforcing intrinsic physical constraints for refinement. After data preparation, event data dimensions such as spatial, temporal, and measurement categories are recovered in a tensor. To prevent overfitting, common in traditional ML methods, tensor decomposition is used. This technique merges intrinsic and physical information across dimensions, yielding informative and compact feature vectors for efficient feature extraction and learning in event detection. Lastly, in grids with insufficient data, knowledge transfer from grids with similar event patterns is a viable solution. This involves optimizing projected and transferred vectors from tensor decomposition to maximize common knowledge utilization across grids. This strategy identifies common features, enhancing the robustness and efficiency of ML event detection models, even in scenarios with limited event data.
ContributorsMa, Zhihao (Author) / Weng, Yang (Thesis advisor) / Wu, Meng (Committee member) / Yu, Hongbin (Committee member) / Matavalam, Amarsagar Reddy Ramapuram (Committee member) / Arizona State University (Publisher)
Created2023