%{search_type} search results

Space exploration drives us to push the boundaries of engineering, especially in the fields of robotics and autonomous systems. Innovations such as satellite communication, GPS technology, and advancements in materials science, originally developed for space missions, have significantly enhanced everyday life on Earth. As we aim to tackle increasingly ambitious space missions -- from the Lunar Gateway to Martian colonization to asteroid mining -- the mobility and manipulation requirements of our robotic platforms exceeds current capabilities. Many of these missions involve environments characterized by adverse gravity conditions and sparse footholds that impede traditional means of locomotion. The solution offered by this thesis, ReachBot, provides unique access to previously inaccessible terrain, broadening our capability to make new scientific discoveries and expanding humanity's reach into space. ReachBot realizes mobility in challenging terrain by combining the portability and maneuverability of small mobile robots with the long reach and high force capability of large robots. It does this by merging two technologies with little previous overlap: mobile manipulation robots and lightweight extendable booms. This thesis begins with a discussion of ReachBot's design, optimized via a trade study method to tailor an extendable-boom architecture to mission-specific parameters and objectives. We demonstrate this method by applying it to a Martian lava tube mission, resulting in a final configuration of booms, joints, and sensors that will provide a continuing example throughout the thesis. Subsequent chapters develop models to describe ReachBot's kinematics and dynamics, including a contact model that addresses uncertainties inherent to grasping in rocky terrain. We then incorporate these models into a robust motion planning algorithm to move ReachBot through its environment while maintaining grasp contact. In particular, we leverage optimal control techniques to generate trajectories that maximize the probability of successful execution. Finally, we validate ReachBot's system capabilities, in addition to our modeling and simulation approaches, with a series of hardware experiments ranging from low-fidelity prototypes using tape measures as deployable booms to a mission-realistic field test in the Mojave Desert. Collectively, this thesis lays the foundation for extensive future work on robotic systems using extendable booms; in particular, the tools used for design, modeling, and control will enable the development of extendable-boom robots built to explore complex adverse gravity environments previously inaccessible to robotic or human exploration.

Emergence and spread of drug resistance against all currently used antimalarial compound classes continues to prevent eradication of malaria, one of the world's deadliest infectious diseases. To aid the effort to discover new antimalarials, we performed a phenotypic chemical screen of compound libraries provided by Medicines for Malaria: Malaria, Pathogen, Pandemic response, and Covid boxes and identified 3 compounds for follow-up. First, a non-bisphosphonate compound, MMV019313 was discovered previously in the lab to target bifunctional P. falciparum farnesyl/geranylgeranyl diphosphate synthase (PfFPPS/GGPPS), a prenyltransferase of the isoprenoid pathway. PfFPPS/GGPPS is a validated, high-priority antimalarial drug target. Unfortunately, current bisphosphonate drugs that inhibit PfFPPS/GGPPS show poor bioavailability and selectivity against the human enzymes. We generated a series of MMV019313 analogs which contained compounds that were potent, selective, and with drastically improved in vitro ADME profiles. Second, we identified a picomolar antimalarial inhibitor, MMV026468, that specifically targets the isoprenoid precursor pathway in blood-stage Plasmodium falciparum. MMV026468's action on the isoprenoid precursor pathway was validated by chemical rescue, microscopy, and western blot. Metabolite analysis demonstrated an overall decrease in isoprenoid precursor pathway intermediates. Additionally, we generated highly resistant, stable, and selective parasites against MMV026468 in three different P.falciparum strains (3D7, W2, Dd2) which underwent whole genome sequencing. Altogether we identified a promising, highly potent, pathway-specific lead compound for development of new antimalarials to combat multidrug resistant parasites. Finally, we identified MMV1580853 (previously BPH-1358), a potent and exceedingly fast-acting blood-stage antimalarial. MMV1580853's rate-of-kill mimics artemisinin in that it kills parasites equally as fast in a 1 hr pulse as treatment for a full 72 hr parasite replication cycle . MMV1580853 has been annotated as a prenyltransferase inhibitor in both humans and bacteria but did not inhibit any of the three annotated P. falciparum prenyltransferases. We did not observed buildup of ubiquitinated proteins, eIF2a phosphorylation, and hemoglobin in MMV1580853-treated parasites, indicating MMV1580853 does not have the same mechanism-of-action as either chloroquine or artemisinin, two fast-acting antimalarials used clinically. MMV1580853 also demonstrated very low mammalian cell cytotoxicity and artemisinin-resistant Kelch13 strains were equally as susceptible to MMV1580853 treatment as non-drug resistant strains. Overall, these properties make MMV150853 extremely favorable as a new lead antimalarial compound for clinical development.

• Early 1919: Reforms to save the nation
• May 4, 1919: Rumors and conspiracy theories
• Late 1919: Marketing with the New Culture Movement
• The 1920s and 1930s: The limits of the new culture movement
• 1919 to 2016: Canonizing a buzzword.

The year 1919 changed Chinese culture radically, but in a way that completely took contemporaries by surprise. At the beginning of the year, even well-informed intellectuals did not anticipate that, for instance, baihua (aprecursor of the modern Chinese language), communism, Hu Shi and Chen Duxiu would become important and famous - all of which was very obvious to them at the end of the year. Elisabeth Forster traces the precise mechanisms behind this transformation on the basis of a rich variety of sources, including newspapers, personal letters, student essays, advertisements, textbooks and diaries. She proposes a new model for cultural change, which puts intellectual marketing at its core. This book retells the story of the New Culture Movement in light of the diversifi ed and decentered picture of Republican China developed in recent scholarship. It is a lively and ironic narrative about cultural change through academic infi ghting, rumors and conspiracy theories, newspaper stories and intellectuals (hell- )bent on selling agendas through powerful buzzwords.

Distributed systems are at the heart of computing today and the distributed protocols stand at the core of these systems. Distributed protocols help to coordinate the state and actions at the different servers constituting the distributed system. These protocols are required to possess two main properties: (i) Correctness—the protocol should correctly execute the action required of it, even in the face of the failure of one or more servers in the distributed system; and (ii) Efficiency—the execution of the protocol should take the least amount of resources in the distributed system as well as the least amount of time. However, despite the past decades of effort that has been spent in designing distributed protocols, achieving both goals (correctness and efficiency) is non-trivial and such protocols are often bottlenecks to many industrial systems (e.g., Kafka, Pulsar, TiDB) in practical use. In trying to understand the reason for these bottlenecks, we realize that, although these protocols aim to satisfy different correctness properties (e.g., linearizability, strict serializability, etc.), these correctness properties all require the protocols to establish certain kind of global ordering agreement across multiple servers, i.e., the multiple servers should take actions according to the same order. The establishment of such a global ordering agreement is not easy for prior protocols, because they have all assumed that the servers of the distributed system do not have a "global view": Each node is only aware of its own state and not the state of other nodes. As a result, prior works usually use one server to play the "leader" role to decide an ordering of the actions; the other ("follower") servers follow the leader's decision and take actions according to the decided order. Such designs can easily cause the throughput bottleneck at the leader, and also cost more latency overheads due to the multiple rounds of communication between the leader and the followers. We aim to avoid such performance bottlenecks caused by the lack of global view. Clock synchronization provides an elegant workaround to do so: By synchronizing the clocks among the servers, these servers are equipped with a global view, or more precisely, a "common clock". After the equipment of the common clock, the servers can refer to their local clock time and take actions simultaneously at a specified time, without relying on a centralized leader to coordinate them. Such a design has not been practical in the past, because tightly synchronizing the clocks at the different servers itself requires a highly performant distributed protocol! Until recently, such a protocol just wasn't available. But, the seminal work of Geng et. al. has developed just such a clock synchronization protocol, making the idea of "common clock" become realistic. Driven by the "common clock" idea, we first develop a general primitive in the thesis, namely deadline-ordered multicast (DOM), which serves as the basic building block for developing high- performance protocols. Simply speaking, DOM works in a "multi-sender-and-multi-receiver" scenario and it equips all the senders and receivers with the common clock. As a result, the sender can specify a future time, called deadline, when it is about to multicast the message to receivers. Receivers, on the other hand, can process the message simultaneously at the specified deadline because they share the same timeline as the sender. Since DOM facilitates the establishment of ordering agreement (on processing messages) among receivers, it helps the distribute protocol to fulfill the correctness requirements more efficiently. Therefore, we are able to develop new distributed protocols which are more performant. Meanwhile, DOM also opens up the possibility of designing new "time-sensitive" systems which were hitherto not possible. In this thesis, we will see some examples of both types. Specifically, Common Clock-based protocols. We will present three distributed protocols, and all of them use DOM to achieve high performance. The three protocols are (1) Nezha (DOM for consensus), (2) Deacon (DOM for concurrency control), and (3) Tiga (DOM for both consensus and concurrency control). New time-sensitive system. We will also present a new financial exchange system called CloudEx. Different from the conventional exchange systems, CloudEx uses "DOM for fairness": By incorporating DOM, CloudEx can create more fair trading environment for the market participants to compete with each other.

The gastrointestinal system is a major organ system that performs numerous functions to maintain organismal health. To maximize our ability to make discoveries about gastrointestinal health and cure gastrointestinal diseases, we must make novel insights into the factors governing gastrointestinal function as well as improve the methods by which we study the gastrointestinal tract. In this dissertation, I present three projects that detail my efforts to do both. In Chapter 1, I summarize the background and motivation for the different projects. In Chapter 2, we examine how the gut physiology of mice, a ubiquitous animal model, is altered by proposed warming of ambient housing temperature in facilities to mouse thermoneutrality (30°C) from standard facility temperatures (22°C). We report, for the first time, a temperature-dependent difference in mouse gut motility that is driven primarily by differences in stress levels experienced at 22°C and 30°C. Additionally, we report the first species-level differences in the gut microbiota of mice raised at 22°C and 30°C. In Chapter 3, we compared the efficacy of different methods for the detection of SARS-CoV-2 in fecal samples from patients with COVID-19. We ultimately recommend a rigorously benchmarked pipeline to maximize the accuracy with which researchers can study the dynamics of fecal SARS-CoV-2 RNA shedding. In Chapter 4, we apply a novel bioinformatic tool to identify insertion-enriched regions of the Salmonella Typhimurium genome that, when disrupted by mobile genetic elements, may provide antimicrobial resistance. We create deletion mutants and screen them against antimicrobials, identifying several that produce altered resistance and sensitivity profiles to numerous chemicals. Finally, in Chapter 5, I summarize the intellectual contributions of this dissertation and discuss recommended future directions for re-search to further the findings from each preceding chapter. Altogether, the work presented here advances both our insights into factors governing gastrointestinal function as well as the methods with which we can study the gastrointestinal tract.

Cell growth and DNA replication are ubiquitous processes of all cellular life. Model bacteria, being genetically tractable and relatively simple to work with, provide excellent systems to investigate cellular growth. Yet, bacterial life is also highly diverse, allowing us to examine a wide variety of questions about how cells grow and proliferate. In this work, we leverage this natural variability to ask two very different questions about cell growth and its relationship to DNA replication. Chapter 2 addresses the question, what are the origins of the fundamental limits of cell growth? We employ the popular model organism Escherichia coli to interrogate the assumption that ribosomes serve as the sole bottleneck of growth rate in bacteria. Using timelapse microscopy of filamenting strains, as well as single molecule tracking of the molecular machines RNA polymerase (RNAP) and ribosomes, this project shows that when genome concentration is diluted by cell growth, growth quickly exits the exponential regime and reaches a plateau. As the genome is diluted, the fraction of active RNAPs and ribosomes decreases, presumably as a result of decreased mRNA concentration. Thus, this project demonstrates that DNA concentration quickly becomes limiting for the attainable cell growth rate, and does so within the physiologic cell size range. This work proposes that bacteria live at the cusp of DNA limitation, a bottleneck of cell growth not previously appreciated. Our findings highlight the economical nature of bacteria and emphasize the importance of initiating DNA replication at a particular mass per origin of replication. Chapter 3 poses the question, how does a bacterium with an unusually complex life cycle grow? Using modified timelapse microscopy techniques, this work shows that the functionally different daughter cells that result from the asymmetric division of the ɑ-proteobacterium Caulobacter crescentus grow at different rates that are associated with the amount of time cells spend in the pre-DNA replication (G1) phase. By following the cell cycle localization of the well-described DNA replication-associated protein MipZ, this project identifies a connection between growth rate and DNA replication not previously appreciated in this bacterium: a requisite increase in cell size that must be added between cell birth and DNA replication initiation, referred to here as the "G1 adder". We speculate on the utility of the G1 adder as a means to integrate growth rate, cell size, and cell cycle progression with the functionality of the G1 phase specific to this bacterium. Finally, this project begins probing the source of the cell cycle correlated growth rate change and proposes areas of future study. Together, these studies represent the questions, from the broad and universal to the fascinatingly specific, that can be examined using bacteria as models of cell growth.

Accurate predictive modeling of gene regulation is crucial for a fundamental understanding of cell identity and function. High-throughput profiling of diverse biochemical and functional properties of cells has enabled powerful deep learning based DNA sequence models that predict protein-DNA binding, chromatin accessibility and histone marks across cell types with state-of-the-art accuracy. Interpretation of these DNA sequence models has revealed novel insights into the cis-regulatory code of TF binding, effects of sequence variation and repeats, and the sequence basis of chromatin accessibility. However, there is much scope for enhancing modeling strategies, model performance, and the tooling and infrastructure around model development and interpretation. Moreover, the full potential of these models for extracting biological insights from high-throughput functional profiling data has not been realized. In this thesis, I will present novel methods that advance DNA sequence models for regulatory genomics, and some applications of these DNA sequence models to glean insights into biological systems. First, I introduce ChromDragoNN, a method that enables generalization of DNA sequence models to make predictions in new cell types. Next, I describe fastISM, an algorithm to significantly speed up variant scoring for convolutional neural networks. I then present dynseq, a tool for sharing and visualization of model-derived importance scores of individual bases. I will then apply DNA sequence models to two different biological systems. First, I combine single-cell chromatin accessibility profiling with DNA sequence models to nominate regulatory DNA variants associated with eye disorders. Next, I apply DNA sequence models to the study of single-cell chromatin accessibility from a time course of human skin cells transforming into induced pluripotent stem cells over four weeks. Using DNA sequence models, I reveal mechanistic insights into reprogramming progression by linking transcription factor abundance changes to sequence logic encoded in regulatory elements. Together, this thesis advances predictive modeling and analysis of gene regulation through new methods, tools and biological applications. I hope that the work moves the field closer to realizing the full potential of DNA sequence models for understanding cell identity and function.

"The Design Card Decks for Ideation and Exploration Web Archive is curated by librarians at the University of Pennsylvania and Princeton University, under the auspices of the Ivy Plus Libraries Confederation. Card decks are used in design fields as a game-based methodology for ideation and stakeholder participatory collaboration. The purposes of decks vary; some are used for brainstorming new products and services while others are scenario-based, prompting designers to consider how their products or services may be incorporated into different communities. They function both as a tool for design speculation and as a set of illustrations that visualize design thinking. This archive preserves websites associated with these decks; as design thinking becomes popular in many disciplines, from business to science, the potential user groups for these tools greatly expands outside traditional design fields. Curators of the archive are: Shannon Robinson (University of Pennsylvania) and Gabriella Karl-Johnson (Princeton University)."

This dissertation explores the development of energy-efficient Deep Neural Network (DNN) accelerators for edge computing, emphasizing real-time inference with strict power constraints. Addressing the shift from cloud-based systems to edge devices, it tackles the embedding of DNN models within devices constrained by memory and processing capacity. The research introduces the Tiny Machine Learning (tinyML) paradigm and custom hardware to support both AI inference and on-device AI training. Focusing on energy-conscious computation in devices powered by limited energy sources like coin-cell batteries, a system architecture is proposed. It balances continuous and event-triggered DNN models to optimize power usage, aiming to operate within an 80-microwatt average power budget to sustain a one-year battery life. The dissertation presents innovative AI device architectures, considering compute elements, memory hierarchies, and dataflows. It details advances in ML algorithms, near-memory computing, and system-level efficiency, addressing data movement bottlenecks. A holistic approach seeks architectures that strike an optimal silicon area and energy efficiency balance, advancing the performance-cost paradigm of tinyML accelerators. The research outlines the design space of tinyML hardware accelerators, the benefits of Non-Volatile memories, custom-designed latch arrays (CLAs), and multicore systems tailored for tinyML. It discusses future directions like larger model integration and heterogeneous memory systems for improved performance. The goal is to facilitate cost-effective, efficient DNN accelerators for small batteries, capable of adapting and learning in their environments. The dissertation introduces CHIMERA, a silicon AI accelerator implemented in 40 nm ULP CMOS, integrating 2 MB on-chip RRAM for DNN weight storage and a RISC-V core. CHIMERA demonstrates exceptional power efficiency, with 126 mW consumption at 1.1 V and 200 MHz, and achieves 2.2 TOPS/W energy efficiency for ResNet-18. It explores power gating and multi-chip performance and proposes a Low-Rank Training algorithm to overcome RRAM's high write energy and latency. To study the trade-offs in hardware accelerators specifically for tinyML applications and achieve sub-microjoule per inference—crucial for persistent sensor data monitoring in IoT—a design space exploration framework named tinyForge has been devised. TinyForge demonstrates the advantages of Custom Latch Arrays (CLAs), which boast 2.7 times higher density and require 5 times less read-access energy compared to traditional synthesized approaches. Leveraging the NSGA-II genetic algorithm, tinyForge conducts a multi-objective optimization to scrutinize the balance between energy, latency, and silicon area, effectively pinpointing Pareto-optimal architectures. This streamlined process culminates in identifying optimal solutions with fewer than a thousand evaluations, significantly expediting the design cycle and leading to time-efficient advancements in tinyML technology. Building upon the insights presented by the tinyForge platform, Medusa stands out as a cutting-edge accelerator tailored for tinyML applications. Medusa integrates the advances in Custom Latch Arrays (CLAs) identified by tinyForge, which offer up to 2.7 times higher density and consume 5 times less read-access energy than conventional synthesized methods. This integration facilitates a significant reduction in memory demands, with Medusa achieving a 4x reduction in memory capacity and a 7x decrease in memory access. Validated on a 28 nm CMOS test-chip, Medusa demonstrates remarkable energy efficiency with less than 0.83 uJ per CIFAR-10 inference, substantially surpassing the performance of existing solutions. The Medusa accelerator's ability to perform full workload execution on edge devices not only confirms its excellence for embedded tinyML tasks but also its preparedness for real-world implementations.