Thesis: S.M., Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2016. / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 57-61). / Road repair expenditure comprises a significant portion of US federal and municipal budgets. Inspection and timely maintenance are crucial preventative measures against pavement distress formation that can lower the monetary burden of repairs. Yet state of the art road inspection techniques still employ technicians to perform distress measurements manually. These methods are often too costly, time-consuming, labor-intensive and require technical expertise. Meanwhile, autonomous systems are increasingly deployed in place of human operators where tasks are monotonous and where risk of exposure to hostile conditions is great. As a time-consuming but highly repetitive task, road inspection presents a promising candidate for task automation. Automating road inspection can present significant efficiency gains that can aid agencies in responding to early signs of erosion in a timely manner. In this work, I explore the capacity of drones to perform autonomous pavement inspections. I develop a system that dispatches drones to survey an area, diagnose the presence of pavement distress in real time, and record imagery and coordinates of locations requiring repair. This system presents an alternative to on-ground inspections and tools that draw on crowd-sourced mechanisms to identify potholes. It builds on other recent technological solutions that employ remote sensing to collect and interpret data on pavement health. The results from this mission will be visualized through a web platform that can not only aid cities in consolidating a fragmented and costly data collection process, but also in minimize human error in the identification and prioritization of problem areas. / by Mia Petkova. / S.M.

Thesis: S.M., Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2017. / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 67-72). / This thesis documents the development of an "expert-free" device in order to realize a system for scalable screening of the eye fundus. The goal of this work is to demonstrate enabling technologies that remove dependence on expert operators and explore the usefulness of this approach in the context of scalable health screening. I will present a system that includes a novel method for eye self-alignment and automatic image analysis and evaluate its effectiveness when applied to a case study of a diabetic retinopathy screening program. This work is inspired by advances in machine learning that makes accessible interactions previously confined to specialized environments and trained users. I will also suggest some new directions for future work based on this expert-free paradigm. / by Tristan Breaden Swedish. / S.M.

Thesis: S.M., Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2018. / Cataloged from PDF version of thesis. / Includes bibliographical references (pages [59]-63). / It's long been known that humans, like many animals, exhibit patterns of behavior that appear to balance exploration of new opportunity and resources with exploitation of already-found safe bets. Humans seem to leverage exploration not only to find quality resources, but also to find quality sources of information, such as people or communities. In this thesis, I explore how exploration behavior and the information diversity afforded by such behavior relates to learning and discovery. I first take a theoretical and algorithmic approach to show how considering exploration behavior and information diversity in deep reinforcement learning systems can lead to improved learning. I then present brief observational studies of exploration behavior in two real-world human systems: a social trading network and human mobility in a major U.S. metro area. In the social trading network, I show that users who fail to seek out diverse information far from their local network are more likely to receive low returns from their portfolios. In the case of human mobility, I find that people tend to have more exploratory relationships with places that are more economically diverse. These studies show that information diversity is closely linked to human exploration behavior, and that inefficient exploration can lead to poorer decision-making. Together, the contributions in this thesis paint a preliminary picture of the importance of information diversity in dynamic networks of learners, be they people or machines. / by Dan Calacci. / S.M.

Thesis: S.M., Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2016. / This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections. / Cataloged from student-submitted PDF version of thesis. / Includes bibliographical references (pages 121-128). / For several decades, secure multiparty computation has been the topic of extensive research, as it enables computing any functionality in a privacy-preserving manner, while ensuring correctness of the outputs. In recent years, the field has seen tremendous progress in terms of efficiency, although most results remained impractical for real applications concerning complex functionalities or significant data. When privacy is not a concern and we are only interested in achieving consensus in a distributed computing environment, the rise of cryptocurrencies, specifically Bitcoin, has presented an efficient and robust solution that exceeds the limits imposed by prior theoretical results. Primarily, Bitcoin's relative efficiency and superiority in achieving consensus is due to its inclusion of incentives. By doing so, it extends the standard cryptographic model to one that reasons about security through rationality of the different players. Inspired by this idea, this thesis focuses on the development of an efficient, general-purpose secure computation platform that relies on blockchain and cryptocurrencies (e.g., Bitcoin) for efficiency and scalability. Similar to how Bitcoin transformed the idea of distributed consensus, the goal in this work is to take secure multi-party computation from the realm of theory to practice. To that end, a formal model of secure computation in an environment of rational players is developed and is used to show how in this framework, efficiency is improved compared to the standard cryptographic model. The second part of this thesis deals with improving secure computation protocols over the integers and fixed-point numbers. The protocols and tools developed are a significant improvement over the current state-of-the-art, with an optimally efficient secure comparison protocol (for up to 64-bit integers) and better asymptotic bounds for fixed-point division. / by Guy Zyskind. / S.M.

Thesis: Ph. D., Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2016. / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 109-118). / Neuronal action potentials ('spikes') are thought to be the fundamental units of information in the brain, hence the ability to record them and to understand their genesis is crucial to our comprehension of the biological underpinnings of our thoughts, memories, and feelings. Over the past several decades an extensive body of work has focused on the mechanisms and timescales over which neurons integrate inputs toward spike threshold. However, most of the work has been carried out in vitro or in silico, and our understanding of what underlies the generation of spike patterns in the awake brain has remained limited. Current models emphasize either seconds-scale global states shared by most neurons in a network, or the fast input integration occurring in single neurons over the few milliseconds preceding spiking, but it's not known whether these represent just the extremes of a continuum. Combining a virtual reality environment with an optimized robotic system for intracellular recordings we therefore analyzed the subthreshold dynamics leading to spiking in a variety of network and behavioral states in the hippocampus, a region known to be involved in spatial navigation, learning and memory, as well as in a model neocortical region, the primary somatosensory cortex. We discovered that the majority of spikes are in fact preceded not only by a fast, monotonic rise in voltage over a few milliseconds, consistent with fast input integration within a neuron, but also by a prolonged, gradual (tens to hundreds of ms) depolarization from baseline, which appeared to exert a gating function on subsequent inputs. Unlike the fast voltage rises, these gradual voltage rises are shared across some, but not all, neurons in the network. We propose that the gradual rises in membrane voltage constitute a novel form of activated state, intermediate both in timescale and in what proportion of neurons participate. By gating a neuron's ability to respond to subsequent inputs, these network-mediated intermediate, or mesoscale, activated states could play a key role in phenomena such as cell ensemble formation, gain modulation and selective attention. / by Giovanni Talei Franzesi. / Ph. D.

Thesis: Ph. D., Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, February 2014. / No formal thesis title-page. Title from abstract page. "September 23, 2013." Cataloged from PDF version of thesis. / Includes bibliographical references (pages 113-119). / Sartorial Robotics is a method of merging fashion theory and robotics through the design and development of robotic systems. These systems facilitate interaction and play as well as mimic the materiality, aesthetics, and construction techniques of textiles, apparel and fashion. This will enhance the social aspects of human-robotic interaction and assist in how we situate robotics in our lives and cultures. Building upon a history of robot aesthetics and a formulaic approach to analyze and understand fashion, a series of design principles for Sartorial Robotics were established and applied in the research and development of robotic systems that utilize the human-centric system of clothing to create robotics for human-robot social interaction. The Group Identity Surface is a soft-architecture system utilizing thermochromic textiles and computer vision to facilitate human-machine teammate building. Zipperbot, a robotic continuous closure for fabric edge joining, was developed to explore autonomous control of a sartorial gesture and performed as a wearable robot which was evaluated through social interactions. Clothing is a uniquely human pursuit and is nearly universal in its adoption and use. It plays a prominent role in our individual cultures transmitting a mixture of social signals and meanings through the semiotics of fashion. It is through this performance of assemblage of fabric surfaces we reconfigure ourselves and our identities. Merging robotics and fashion within the practice of Sartorial Robotics will enhance the explorations of identities for both humans and robots. / Adam Whiton. / Ph. D.

Thesis: S.M., Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2016. / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 76-77). / Budgeting and expenditure data is the clearest expression of a government's priorities. Despite its importance, making it available to the public imposes hard challenges that not every administration is ready to undertake. The lack of IT capabilities and constrained resources of government agencies -regardless of size or budget- make it difficult for them to respond to the demands of information from their constituencies, transparency advocates, the press and central governments. Moreover, these administrations don't usually have access to analysis tools that help them view how the resources are being used and to detect potential risks of misspending -a critical need of elected officials, who are under everyday scrutiny. Responding to these needs, I propose the design, implementation, impact and usability studies of a software platform for analysis and visualization of government budget and expenditure data. / by Manuel Aristarán. / S.M.

Thesis: Ph. D., Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2018. / Cataloged from PDF version of thesis. Vita. / Includes bibliographical references (pages 179-201). / Today's educational systems are complex, political, sociotechnical ecosystems that struggle to meet the needs of most learners and societal demands-and most critically, struggle to change. Yet, learners globally need access to high quality learning environments and coherent learning pathways that support them to thrive in our complex world. Fundamental to every learning technology, environment, and system, is a learning data model and architecture that helps to organize the learner's experience. To date, in traditional educational systems, this has largely been dominated by public policy curriculum standards, which have tremendous limitations and shortcomings on classroom practice and their ability to support complex learning technologies. At the same time, over the past several decades significant advances have been made in the learning sciences, learning analytics, and learning technologies that have greatly expanded our ability to model learning and provide immersive and adaptive learning environments. Yet each of these communities rarely coordinates and aligns these data models. The disjointedness of these structures leaves their architecture in a messy, challenging state, unable to successfully carry us into an advanced future of learning technologies and effective learning ecosystems. This dissertation explores the use of Systems Engineering as a means to reengineering this critical infrastructure of the system, through the LearningGraph-a research initiative that used this methodology to create a unified data structure for modeling learning constructs in a coherent learning data architecture. The aim of the project is to ultimately inform a new infrastructure to support learning development across learning technologies and environments. In doing so, we create the foundation for closing significant gaps in the current system: between learning sciences research and practice; curriculum and assessment design; the design of learning technologies and all the aforementioned components; and between and across education systems globally. Moreover, it creates the potential for the foundation of a very different future for learning ecosystems. / by Jennifer Sterling Groff. / Ph. D.

Thesis: Ph. D., Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2017. / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 258-274). / This thesis is about designing shape change composite material for interactions. Interaction has gone beyond computer screens and electronics to enter the realm of physical materials. Shape changes at the micro level will cause shape changes and other physical property changes at the macro level. A design strategy for bioinspired shape-changing composite materials includes two development steps: a shape-changing material unit (SCMUnit), followed by a shape-changing matrix composite (SCMC). SCMC contains the matrix phase and the dispersion phase, one of which is composed of SCMUnits. In addition, SCMC can be hierarchical, while SCMC and SCMUnits have a relationship of recursive embodiment. Two major projects exemplify how water-responsive shape-changing material can be used to design interactions based on the outlined design strategy. bioLogic is about hygromorphic bacteria-based SCMC, while Transformative Appetite is about water-driven edible SCMC material. Programmable transformations, multilayer composites and sequential-foldings were engineered with these materials. A customized fabrication strategy, combining wet lab processes and additive manufacturing, was introduced, while applications were presented to exemplify various interaction scenarios. In addition, the SCMC design strategy has been adapted to develop shape-changing materials beyond water responses. Stimuli and responsive behaviors are used to categorize these materials. A design space for nature-inspired responsive material design for shape-changing interfaces was outlined from two aspects: the technical aspects and the conceptual aspects. The technical aspects are identified with the interplays of three features of nature: natural structural mechanisms, natural stimuli and natural transformation mechanisms. The conceptual aspects is summarized in two conceptual spaces: microscale shape changes for macroscale shape changes, and microscale shape changes for macroscale material property changes. / by Lining Yao. / Ph. D.

Thesis: S.M., Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2016. / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 72-73). / In this thesis we present a novel approach to the computer simulation of forward dynamic gait models and the optimization of parameters that must be tuned for such models. This methodology is not limited to gait simulation, and could be useful for any situation in which a complex Simulink model requires variables to be tuned via machine learning to optimize all heuristic that can only be evaluated via simulation. Through the lens of Biomechatronic engineering research, we combine the fundamentals of software engineering with a refinement of the best practices of Matlab and Simulink programming and a working knowledge of inherent Matlab and Simulink constraints to construct a framework for rapid model development. Key features of this methodology include: the use of Simulink as an environment for rapidly prototyped models, the use of and construction of custom Simulink libraries, and use of the Matlab Optimization Toolbox. This methodology uses parallel evaluation of rapid acceleration Simulink executables to minimize optimization time, and allow research teams to take advantage of parallel processing and cloud computing. This methodology was applied to a bouncing gait model developed by Hartmut Geyer for evaluation. We demonstrate its effectiveness by simulating this model using a custom library of model components, such as ground contact model, Stateflow control, heuristic computation, and body segments. / by Matthew D. Furtney. / S.M.