12023 research outputs found
Sort by
Robust Safety-Critical Control: A Lyapunov and Barrier Approach
Get PDFAccompanying the technological advances of the past decade has been the promise for widespread growth of autonomous systems into nearly all domains of human society, including manufacturing, transportation, and healthcare. At the same time, there have been several tragic failures that reveal potential risks with the expansion of autonomous systems into everyday life, and indicate that it is vital for safety to be accounted for in the design of control systems.
This thesis seeks to develop a theory of robust safety-critical control for autonomous systems. This theory will be built upon the foundational tools of Control Lyapunov Functions (CLFs) and Control Barrier Functions (CBFs), which provide a powerful paradigm for the design of model-based safety-critical controllers. The dependence of CLF and CBF-based controllers on a system model makes them susceptible to modeling inaccuracies, potentially resulting in unsafe behavior when deploying these controllers on real-world systems.
In this thesis I present methods for resolving four classes of model inaccuracies referred to as model error, disturbances, measurement error, and input sampling, which are commonly faced challenges when designing controllers for robotic systems. The proposed methods are unified by their shared use of CLFs and CBFs to produce controllers possessing rigorous and robust safety guarantees that can be demonstrated in simulation or experimentally. A hallmark of these methods is a focus on enabling control synthesis through convex optimization, which ensures that controllers can be efficiently computed on real-world robotic hardware platforms.
In addressing model error, I consider both data-driven learning approaches and adaptive control approaches. I present three episodic learning frameworks that iteratively augment existing CLF and CBF-based controllers specified via convex optimization problems to improve the stability and safety properties of a system, which I demonstrate in simulation and experimentally. I also establish a relationship between the degradation of stability and safety properties with the magnitude of residual learning error through the perspective of Input-to-State Stability (ISS) and Input-to-State Safety (ISSf). Lastly, I develop an adaptive safety-critical control framework for systems with parametric model error through the notion of adaptive CBFs.
In addressing disturbances, I resolve challenges in balancing performance and robustness with ISSf-based controllers through the notion of Tunable Input-to-State Safety (TISSf), which permits prioritizing robustness to disturbances only when safety requirements are close to being violated. I demonstrate the capabilities of TISSf-based control design experimentally on an autonomous semi-trailer truck system that is subject to input disturbances due to complex unmodeled actuator dynamics. Lastly, I develop a framework for achieving ISSf-like finite-time safety guarantees for discrete-time systems subject to stochastic disturbances through the use of CBFs and convex optimization.
In addressing measurement error, I develop the notion of Measurement-Robust CBFs (MR-CBFs), which permit control synthesis through convex optimization in the presence of imperfect measurements. I demonstrate the capability of MR-CBFs on an experimental Segway system using a vision-based measurement system, validating the tractability of using controllers specified through increasingly complex classes of convex optimization problems on real-world systems. Lastly, I present an application of Preference Based Learning (PBL) in tuning the robustness parameters of a CBF-based controller, demonstrating the first use of PBL with CBFs and providing a tool for tuning the safety and performance of the robust controllers proposed in this thesis.
In addressing input sampling, I consider both sampled-data and event-triggered paradigms for modeling input sampling. I provide a method for synthesizing CLF-based controllers for sampled-data systems by integrating feedback linearization with approximate discrete-time models, leading to a significant improvement over continuous-time CLF-based controllers implemented with input sampling. I then develop a framework for achieving safety of sampled-data systems through approximate discrete-time models through the notion of practical safety and Sampled-Data CBFs (SD-CBFs), which I demonstrate with convex-optimization based controllers in simulation. Lastly, I develop a method for event-triggered safety-critical control that uses ISSf to achieve safety while satisfying the requirement of a minimum interevent time.
Collectively, these contributions constitute a significant advance in the theory of robust safety-critical control by establishing a framework, unified by the use of CLFs and CBFs in conjunction with convex optimization, that addresses a wide class of challenges faced in the design of safety-critical control systems.</p
Enantioselective Syntheses of Tetrahydroisoquinolines (THIQs) via Iridium-Catalyzed Asymmetric Hydrogenation and Progress Toward the Total Synthesis of (+)-Cyanocycline A
Get PDFDescribed herein are two reviews and three projects related to the asymmetric syntheses of tetrahydroisoquinoline (THIQs) alkaloids, and the progress toward the total synthesis of (+)-cyanocycline A. In Chapter 1, a review of the development of asymmetric methodologies for the preparation of enantioenriched N-heteroarenes is detailed. In Chapter 2, the development of an iridium-catalyzed enantio- and diastereoselective hydrogenation of 1,3-disubstituted isoquinolines to achieve cis-THIQs is reported. Chapter 3 describes the iridium-catalyzed asymmetric hydrogenation of 1,3-disubstituted isoquinolines that can afford trans-THIQs in a single transformation. Preliminary mechanistic insights to the iridium-catalyzed asymmetric hydrogenation method using deuterium experiments are detailed.
Chapter 4 details a comprehensive review of the advances in the total syntheses of complex THIQ alkaloids from 2000 – 2020, ranging from simple benzyl THIQ natural products to complex THIQ alkaloids such as Ecteinascidin-743. In Chapter 5, efforts toward the total synthesis of (+)-cyanocycline A are described, harnessing a non- biomimetic synthetic route through a convergent cross-coupling of two heterocyclic fragments followed by a global hydrogenation event.</p
Design Rules for Multi-Electron Systems in Next-Generation Batteries: From Mg Electrode-Electrolyte Interface to Anion Redox Activation in Li-Rich Sulfides
No full textLi-ion batteries (LIBs) have revolutionized the modern world, powering portable electronic devices and more recently realizing electrification of transportation. With more technological advancements that further improved the performance, LIBs also play an important role as one of the most promising energy storage systems in transforming into renewable energy sources and achieving net zero emissions. However, state-of-the-art intercalation-based LIBs are beginning to mature and reach their theoretical capacity limits. To further improve the electrochemical performance of batteries and meet growing demands of energy storage applications, there have been growing efforts to increase the energy density beyond the limits of conventional LIBs. In this thesis, we examine two examples of multi-electron systems–Mg electrolytes and Li-rich sulfide cathode materials–to gain insights and establish design principles.
First, we explore the magnesium aluminum chloride complex (MACC) electrolyte to study the role of the electrode-electrolyte interface in Mg charge transfer. We demonstrate that MACC electrolyte which normally requires electrolytic conditioning can be chemically activated by the small addition of Mg(HMDS)₂. Solution-phase characterization reveals that Mg(HMDS)₂ helps prevent the formation of passivating film on the Mg surface by scavenging trace amounts of H₂O. Mg(HMDS)₂ also reacts with MACC to form free Cl⁻ which decorates the Mg surface which facilitates Mg electrodeposition and stripping.
Next, we investigate three different alkali-rich sulfides-LiNaFeS₂, LiNaCoS₂, and Li1.33-1.33zTi0.67+0.33zVaczS₂ - to probe the role of electronic and physical structure in governing reversible anion redox. We demonstrate that cryomilling LiNaFeS₂ mitigates particle fracturing by increasing microstrain and reducing crystallite size. Isostructural LiNaCoS₂ exhibits more covalent interactions between the transition metal-d and S-p states compared to LiNaFeS₂, but undergoes an irreversible conversion reaction. Lastly, Li₂TiS₃ exhibits no electrochemical activity, but introducing cationic vacancies in Li1.33-1.33zTi0.67+0.33zVaczS₂ activates S oxidation. Li1.33-1.33zTi0.67+0.33zVaczS₂ is studied further to study first-cycle activation and voltage hysteresis in Li-rich sulfides.</p
Mechanism of Response and Resistance to CAR T Cell Therapies
No full textWhile CAR T cell therapy has demonstrated remarkable success in treating leukemia, lymphoma, and multiple myeloma, its effectiveness in treating solid tumors, such as glioblastoma (GBM), remains limited. It is imperative to comprehend the mechanisms that hinder the efficacy of CAR T cell therapies and to devise strategies to counteract tumor resistance. In this study, we hypothesized that disruptions in the cell-intrinsic interferon (IFN) signaling pathway contribute to the establishment of an immunosuppressive tumor microenvironment in solid tumors, ultimately rendering solid tumor cells resistant to CAR T cell-mediated elimination.
To investigate this, we established syngeneic models of IFN signaling-deficient tumors in the context of murine IL-13Ra2 targeted CAR T cell therapy. Our findings revealed that these models indeed manipulate the tumor microenvironment (TME), leading to a resistance to CAR T cell therapy. Notably, we observed variations in gene expression related to IFN signaling components and cytokines between IFN signaling-deficient tumor cells and wild type (WT) tumor cells following CAR T cell treatment.
Moreover, employing techniques such as single-cell RNA sequencing and mass cytometry analysis, we scrutinized the immune cell infiltrates within tumors lacking IFN signaling in comparison to WT controls. This in-depth analysis identified key immune-mediated factors contributing to the resistance observed in tumors with disrupted Janus Kinase1 (JAK1/KO) signaling upon CAR T cell therapy. Specifically, CAR T cell-treated IFN signaling-deficient tumors exhibited reduced T-cell transcripts, characterized by a decline in the frequency of CD8-early active and CD8-native like T cells. Conversely, an increase in regulatory and follicular T cells, exhausted endogenous T cells, and even exhausted CAR T cells was observed in treated IFN signaling-deficient tumors when contrasted with treated WT tumors. Furthermore, our analyses underscored the enrichment of genes associated with fibroblasts, neutrophils, and myeloid cells within IFN signaling-deficient tumors in contrast to WT tumors.
These findings collectively suggest that the immune suppressive communication within the IFN signaling-deficient tumor microenvironment could arise due to the differential enhancement of receptor-ligand interactions, such as SPP1+ tumor-associated macrophages (TAMs) and CD44+ cancer-associated fibroblasts (CAFs), along with interactions involving integrins on other cell lineages. To address the resistance observed in IFN-deficient tumors, we devised strategies aimed at enhancing the efficacy of CAR T cell therapy by promoting the recruitment of activated immune cells and reshaping the tumor microenvironment.
We next conducted an analysis of immune signatures in 32 GBM patients who exhibited progressive disease following CAR T cell treatment, comparing them with patients who displayed relatively stable disease or showed signs of improvement. Our investigations revealed the presence of fibroblasts and SPP1+ APOE+ C1QA+ C1QC+ myeloid cells within the GBM signatures, which are associated with immune suppression and resistance to therapy. Notably, patients with GBM who displayed a relatively stable treatment response and enhanced T cell recruitment demonstrated distinct expression patterns of interferon regulatory factors (IRFs) compared to patients with less favorable treatment responses.
These findings offer insights into the intricate interplay between tumor-intrinsic driver mutations, the composition of the tumor microenvironment, and the responsiveness of solid tumors to CAR T cell therapy. Importantly, our study provides potential avenues for addressing resistance in tumors that do not respond to interferon-based therapies.</p
Geological and Geochemical Explorations of the Salitre Formation Phosphorite, Eastern Brazil
Get PDFThis thesis documents my explorations of an ancient seafloor environment through sedimentary geology and stable isotope geochemistry. The geologic record of this seafloor — its life, environmental conditions, lithification, burial and exposure — consists of hundreds of meters of sedimentary rock, outcropping across Bahia and Minas Gerais, Brazil. Though it consists primarily of carbonate grains surrounded by carbonate cements, such as might be found forming in any shallow carbonate platform, this record also contains one of Brazil’s most extensive sedimentary phosphate deposits. In these deposits, phosphate is concentrated as carbonate fluorapatite cements (CFA) in digitate stromatolites, distinctively finger-like, branching accretionary structures likely formed by the accumulation of sediment by microbial mats and biofilms. Chapter 1 introduces the broader motivations of this kind of paleoenvironmental and paleoecological research, for understanding the record of life on Earth and other worlds. Chapter 2 presents new sedimentological and stratigraphic data which interpret the depositional setting of a seafloor 600 million years ago on the rending supercontinent of Gondwana. Chapter 3 presents new carbon and oxygen isotopic measurements and clumped isotope measurements of structural carbonate in phosphatic and non-phosphatic textures of the rock, and uses them to constrain the alteration history of the rock and its effect on the record of primary depositional conditions. Chapter 4 presents new data on the sulfur isotope composition of specific minerals in the rock, combining several disparate analytical methods to draw conclusions about the metabolism of the stromatolites’ microbial architects. Chapter 5 describes the distribution of organic material and style of fossilization, and presents preliminary data which suggest a possible mechanism by which microbial activity may have facilitated phosphate concentration and mineralization. As a whole, this thesis demonstrates the value of multidisciplinary analyses in the reconstruction and understanding of sedimentary phosphorite deposits throughout Earth history, improving our understanding of how and to what extent phosphorites may record the history of life and the environment
Coordination Between Mammalian Nascent Protein Targeting and Cotranslational Chaperones
Get PDFProtein biogenesis starts with ribosome synthesizing nascent polypeptide chain. Ribosome is a major hub for multiple pathways including membrane targeting, chaperones, chemical modification, and quality control. All these pathways need to coordinate with each other spatially on the ribosomal surface and temporally within the translation elongation window. Accumulating data start to point to more intricate interaction and coordination between different pathways beyond the simple competition traditionally presumed.
In Chapter 1, I demonstrate the coordination between a cotranslational chaperone, NAC, and the ER targeting machinery, SRP. NAC and SRP can bind to the same ribosome simultaneously despite overlapping binding sites, allowing NAC to change conformation of SRP specifically to the NC sequences. This allostery enhances the specificity of SRP-SR association, explaining the long-observed effect of NAC modulating ER targeting specificity.
In Chapter 2, I dig deeper into the mechanism of NAC regulating SRP. Based on cryo-EM structures, NAC domain sits on top of the ribosomal tunnel exit, potentially sensing the identity of NC, and is anchored by positively charged NACβ N-terminal tail. NAC-UBA domain is the key to recruiting SRP and coordinating the substrate handover to SRP.
In Chapter 3, I focus on the cotranslational HSP40/HSP70 system of RAC. Ribosome binding of RAC stimulates its cochaperone activity to activate HSP70 ATP hydrolysis. Ribosome sensing by RAC is related to the NBD of HSPA14. RAC-stimulated HSP70 engagement to NC keeps it in a folding-competent unfolded state before HSP70 dissociation.
Taken together, this study advances the experimental and theoretical tools to studying cotranslational pathways associated with the mammalian ribosome and demonstrates the interesting question of coordination between cotranslational pathways.</p
Exploring the Mass Accretion Rates of Neutron Star X-Ray Binaries and the Properties of Cadmium Zinc Telluride for Hard X-Ray Astronomy
Get PDFMy thesis work has consisted of investigations of several X-ray binaries using the Nuclear Spectroscopic Telescope Array (NuSTAR), as well as characterization of NuSTAR-like hybrid X-ray detectors for the purpose of future hard X-ray obser- vatories. I present analyses of three X-ray binaries in particular, probing different luminosities with the goal of investigating accreting neutron stars at different ac- cretion rates. At high accretion rates, approaching and moderately exceeding the Eddington limit, I performed a timing and spectral analysis of the neutron star su- pergiant X-ray binary SMC X-1. I tracked the transient pulsations of this source at different apparent luminiosities and analyzed the source spectrum as pulsations evolved in order to infer that the mechanism leading to pulsation transience is likely related to periodic obscuration by a warped accretion disk. At lower accretion rates, I studied the low-mass X-ray binary GRS 1741.9–2853. I proposed for NuSTAR observations of this source in outburst, and as a result I observed two type-I X- ray bursts originating from the source. I performed a time-resolved spectroscopic study of each of these bursts, and I used the results of this analysis to determine the composition of the burning material and the distance to the source from Earth. The third X-ray binary I present is MAXI J1848–015. This source was discovered by the Monitor of All-sky X-ray Image (MAXI) in winter 2020 at which point its luminosity was abnormally low for an accreting compact object in outburst, making it a good candidate for the study of neutron star X-ray binaries at low accretion rates. I performed rapid follow-up of the source with NuSTAR, and I performed a detailed spectroscopic study of the NuSTAR observations, utilizing relativistic disk reflec- tion models in order to determine that the source is in fact a black hole rather than a neutron star. Complementing these observational astrophysics projects, I character- ized Cadmium Zinc Telluride (CZT) detectors procured from Redlen Technologies in order to determine their usefulness for future hard X-ray observatories. Using Python, I produced an adaptable calibration and analysis pipeline with which to analyze noise, leakage current, spectral, and other data which I obtained in the lab. Using this pipeline, I determined that the material available from Redlen shows good uniformity, yield, and superior spectral resolution up to high photon energies, making it an excellent candidate for application to hard X-ray astronomy.</p
Applications of Genetically Engineered Bacillus subtilis in Biocatalysis and Functional Materials
Get PDFBacillus subtilis is a gram-positive model bacterium that forms endospores as a response to nutrient limitation and other environmental stresses. The B. subtilis spore contains a dehydrated core, where the bacterial genome is safely stored, and multilayer proteinaceous coats, protecting the spore from various physical and chemical insults. Because of the outstanding resilience of the B. subtilis spore, it has attracted increasing interest for application in biotechnology. In this thesis, we demonstrate the utilization of genetically engineered B. subtilis cells and spores for heterologous protein display and functional material synthesis and characterization.
In Chapter 1, we review the fundamentals of sporulation and germination in B. subtilis. We highlight notable biotechnological applications of native and engineered B. subtilis spores in recent years. We also discuss limitations associated with prior studies that inspire us to pursue the work in this thesis.
In Chapter 2, we describe the T7 RNA polymerase (RNAP) enabled high density protein display on B. subtilis spores (TIED) method. The TIED constructs employ a coat protein promoter – PcotG, PcotV, or PcotZ – to drive the expression of the T7 RNAP. Target proteins are fused to the C-terminus of a spore crust protein – CotY or CotZ – and subjected to amplification by the T7 promoter. We prepare the endogenous constructs in which coat protein promoters directly regulate fusion protein expression for comparison with TIED. In addition, we develop a supplementary procedure to harvest spores before mother cell lysis, further improving the loading density of the target proteins. We verify the performance of the TIED architectures with a fluorescent reporter protein, mWasabi. Together with the early harvest protocol, the TIED method substantially enhances the total expression level and loading density of the crust-mWasabi fusion proteins relative to the endogenous expression system, as evidenced by bulk fluorescence measurements and microscopy.
In Chapter 3, we implement the TIED architectures described in Chapter 2 for enzyme display on B. subtilis spores. We demonstrate the spore-based biocatalyst platform with three enzymes – lipase A and lipase B secreted by vegetative B. subtilis, and an engineered peroxidase, APEX2. We manifest that TIED enables massive accumulation of all three enzymes on the spore surface, with loading densities in the range of 106-107 enzymes per spore. Further, TIED-enzymes show comparable catalytic performance to the respective free-form enzymes, enhanced catalytic activity in methanol, and increased temperature stability. We conduct Michaelis-Menten studies to elucidate the kinetic characteristics of TIED-enzymes and their free form counterparts. Finally, we demonstrate that TIED-enzymes are not only recyclable, but also fully renewable after loss of activity through induction of germination and sporulation, demonstrating the potential for perpetual regeneration of the immobilized biocatalysts.
In Chapter 4, we describe a new class of living composite materials (LCMs), in which genetically engineered B. subtilis cells and spores are effectively crosslinked into the surrounding polymeric scaffold. The resulting LCMs can be dried to yield portable materials. When re-immersed in aqueous media, entrapped cells and spores in previously- dried LCMs exhibit metabolic activity, including synthesis and secretion of recombinant proteins. Notably, we show that the scaffold based on photopolymerization of N-(hydroxymethyl) acrylamide (NHMAA) achieves effective cellular confinement, showing no evidence of cellular leakage over a period of 72 hours. We envision that the design principles elucidated in this work can provide a promising route to functional living materials engineered for biomedical and other applications.</p
Investigating the Earthquake Cycle on Multiple Temporal and Spatial Scales Using Satellites and Simulations
Get PDFThe motion of the Earth's tectonic plates creates a gradual accumulation of stress at their boundaries, followed by a rapid release in earthquakes, a process known as the earthquake cycle. Studying this process is important because of the hazards earthquakes pose, but presents challenges due to the multi-scale nature of the problem—stresses build up over hundreds to thousands of years, while earthquakes break narrow fault zones in a matter of seconds. In this thesis, we combine a variety of techniques to study the earthquake cycle on multiple temporal and spatial scales, including satellite-based interferometric synthetic aperture radar (InSAR) to observe the slow deformation of the Earth over wide areas, and high-performance computational simulations to model faults during earthquakes. We begin by presenting a method for removing the signal of plate-tectonic motion in large-scale InSAR measurements, allowing for better observation of small ground deformations. We then use these corrections to study the Makran subduction zone, on the Iran-Pakistan border. Our InSAR-derived ground velocity map can resolve motions at the level of millimeters per year over an area of nearly one million square kilometers, and we use it to place constraints on the degree of coupling on the subduction megathrust. Next, we show how InSAR can be combined with deep learning techniques to rapidly map earthquake damage in all weather conditions, day and night. Such products will hopefully prove useful in future disaster response. Finally, we present computational simulations of dynamic earthquake ruptures with enhanced dynamic weakening due to thermal pressurization. We apply our simplified model to the creeping section of the San Andreas Fault, which is generally thought to be a barrier to earthquake rupture. Our results show how thermal pressurization can allow earthquakes to propagate partially or completely through the creeping section for a range of physically reasonable parameters. Our work illustrates how results from multiple fields can be combined to deliver new insights into the earthquake cycle and the hazards that it poses
Redox-Activated Covalent Functionalization of Semiconductor Surfaces
Get PDFSemiconducting materials are those with a band gap across which electrons can be excited when the material absorbs photons with sufficient energy. Surface functionalization of semiconductors involves manipulation of the properties of the material by attaching organic small molecules through a surficial covalent bond. By controlling the surface properties of the material, functionalization has enabled the application of semiconductors in a myriad of fields, prompting a highly active field of research. To aid in this effort, we explore a new reaction methodology based on redox-mediated surface functionalization, where an outer-sphere, one-electron metallocene reductant or oxidant is added to the solution medium containing the semiconductor and the small molecule to be added to its surface. Using density functional theory, we elucidated the thermodynamic and kinetic factors that limit the experimentally observed upper coverage bound of reductant-activated methylation of 1T′-molybdenum disulfide by determining two governing factors: 1) sulfur sites with longer Mo–S bonds are more thermodynamically favorable for methyl addition, and 2) sulfur sites with fewer adjacent methylated sulfur sites are preferentially functionalized due to steric hindrance. We then expanded this reductant-activated reaction methodology to silicon(111) surfaces and demonstrated that the reductant solution potential must lie near or above the silicon(111) conduction band edge to observe reactivity. By extending this study to silicon nanocrystals of different sized diameters and different conduction band edges, we found that the extent of surface reactivity relied heavily on reductant strength, but the energy difference between the conduction band edges was too small to observe a distinct dependency on nanocrystal size. The work encompassed in this thesis expanded our understanding of redox-mediated reactions on semiconductor surfaces, providing a new avenue for attaining atomic-level control of the surficial properties of the material using mild reaction conditions and no specialized equipment. Furthermore, redox-activated addition enables the use of new functional groups that would otherwise be reactive in other functionalization methods, promoting an abundance of opportunities to explore new applications of semiconductor materials