150813 research outputs found
Sort by
Choosing Networks for Ride-Hailing Platforms
No full textThe development of autonomous vehicles is poised to reshape the landscape of transportation. As companies prepare to deploy these vehicles on ride-hailing platforms, a key operational challenge is determining the networks on which to train the vehicles. Our work contributes toward addressing this challenge on three fronts. First, we develop a theoretical model of the network selection problem and prove theoretical results that show the importance of two parameters: the detour factor and the fleet size. Second, we develop several approaches for selecting the networks. Third, we evaluate these approaches on empirical data. We find empirical support for the importance of the detour factor and the fleet size.M.Eng
On the use of high‐density polyethylene bottles for long‐term storage of total alkalinity samples
No full textTotal alkalinity (TA) plays an important role in buffering seawater and determining how much anthropogeniccarbon dioxide the oceans can absorb and mitigate the rise in atmospheric concentrations. Total alkalinity varieswith location, depth, and time making it an important variable needed to quantify and monitor ocean acidification,and potentially for ocean alkalinity enhancement interventions. Currently, best practices are to use expensivehigh-quality borosilicate glass bottles for collecting and storing these samples. However, unlike other carbon systemvariables, TA is not affected by gas exchange meaning plastic bottles may be suitable for TA sample storage. Plasticbottles are lighter, cheaper, and less prone to breakage making them easier to handle and ship. Here, we test the suit-ability of high-density polyethylene (HDPE) for collection and long-term storage of TA samples. In two sets of exper-iments, it was determined that HDPE is not suitable for long-term storage of TA samples as there were large changesin TA over time and precision of duplicate samples was very poor. We hypothesize that HDPE plastic is slightlyporous leading to leaching of alkalinity either into or out of the bottle over time impacting the value of the sample.Use of HDPE bottles for TA samples is not recommended for long term sample storage
Space-Based Solar Power: Implications for Operational Robustness in Lunar EVAs and Exploration Architectures
No full textHuman exploration of the lunar surface has large power requirements for both the lunar base and for rover exploration. NASA’s recent contract awards indicate a reliance on fission surface power. While nuclear options provide reliable power to lunar base locations, they have a limited reach that restricts exploration capacity. The Space Exploration Vehicle’s 125-mile range only allows coverage of 0.34% of the lunar surface. A constellation of space-based solar power (SBSP) satellites paired with pressurized rovers allows 24-h, full-surface coverage on excursions from the lunar base. A case study is conducted of the constellation design, system cost, operational lifetime, and power provided using SBSP. Results of the case study demonstrate that SBSP provides an additional 20 kW/h of emergency power and extends EVA range from 125 to 1000 km to cover 26 of the lunar geologic units, at an added lifecycle cost of less than 1% of the baseline mission cost. Addition of a SBSP constellation for rovers provides operational flexibility, safety, and robustness to enable multiple lunar exploration architectures beyond that enabled by surface power infrastructures, and should be further explored for lunar missions
From Sketch to Stage: Tools for Prototyping and Exporting Collaborative DMIs on the Web
No full textThis thesis presents tools and ideas for prototyping and exporting collaborative digital music instruments (DMIs) on the web, the primary purpose of which is to lower the barrier to making music and to enable easier collaboration. This is done in the context of the Creativitas website, which has become a tool of the MIT 21M.080 "Introduction to Music Technology" class to learn about music technology and audio on the web, and a tool for FaMLE (the Fabulous MIT Laptop Ensemble) to use in live performances. The website allows creators to execute code within an editor code box and partake in a practice known as live coding, ultimately creating both sound and visuals. Audio is primarily created with the Tone.js interactive web audio framework, and visuals are drawn on a provided canvas using p5.js. This thesis extends the Creativitas website by providing functionality for exporting the written code as a standalone website. The exported standalone websites serve as DMIs, with standard controls such as volume, tempo, and start and stop buttons. Furthermore, we discuss and implement strategies for synchronizing timing and instrument values. This includes state-of-the-art strategies, as well as ideas for creating extendable interfaces that can include more strategies as they are developed. We end with two examples of exported DMIs, which can be effectively used in performances.M.Eng
She Swims in Silence: Spatial Narrative, Women's labor in Contemporary Art
No full textThis thesis investigates the collective lives of Chinese women sent to Xinjiang in state-led migration after 1949 and the erasure of their gendered narratives. Drawing on a unique family history and archival evidence, the thesis reveals how the personal identities of these female “Aid to Xinjiang” participants were stripped away and subsumed under the grand socialist nation-building myth. Through practice-based artistic research, the project attempts to restore their lost voices and unacknowledged suffering and labor, framing the exhibition as a form of praxis. By analyzing the exhibition alongside case studies and critical analysis, the thesis, inspired by Bernard Stiegler’s theory of the “history of representational forms” and interwoven with ideas from philosophers like Judith Butler and Nicholas Mirzoeff, interrogates the gendered silences in official history and highlights the tension between state mythologies and personal memories. In doing so, the exhibition as an interdisciplinary form of research not only restores agency to a silenced group of women, but also demonstrates how artistic practice can serve as an alternative historiography to challenge dominant narratives and recover marginalized voices.S.M
Design and Fabrication of Hybrid Functional Identities for Mechanical Elements
No full textCHI EA ’25, Yokohama, JapanMy PhD research explores the simultaneous integration of mechanical and electrical functionalities in mechanical components such as gears, linkages, and springs, which I define as "hybrid functional identities." The focus is on transforming these components into non-intrusive sensors and active elements that maintain structural integrity while providing electrical capabilities like sensing, energy harvesting, and communication. I establish a framework for hybrid functional identities by examining common mechanical elements and their associated motions—rotational, linear, and reciprocal—along with force-based interactions like stretching, compression, and torsion. This analysis identifies essential electrical functionalities that complement these mechanical behaviors. Building on this foundation, I investigate modular mechanical building blocks that support diverse mechanical and electrical interaction primitives using a unified geometric structure. Ultimately, I aim to create an interconnected system where hybrid mechanical-electrical components function autonomously and communicate through an embedded wireless network
Efficient Generative Models for Visual Synthesis
No full textWhile current visual generative models produce high-quality outputs, they suffer from significant computational costs and latency, limiting their applicability in interactive settings. In this dissertation, we introduce a suite of techniques designed to enhance the efficiency of generative models for image and video synthesis. First, we propose distribution matching distillation, a method that enables the training of one- or few-step visual generators by distilling knowledge from computationally expensive yet highly capable diffusion models. Next, we develop improved distillation techniques that enhance robustness and scalability, culminating in a production-grade few-step image generator. This system is now deployed in widely used software, generating hundreds of millions of images annually. Finally, we extend our approach to video generation by adopting an autoregressive paradigm, significantly reducing latency and enabling fast interactive video generation and world simulation.Ph.D
Mantis: A Screen Magnification Tool for Diagram Traversal
No full textComplex diagrams and charts can be difficult for people who use screen magnification to navigate. A sense of spatial context and of the diagram’s overall structure is oftentimes lost, as magnifiers can only magnify a fraction of the screen at any given time. So, while sighted users have both clarity and full context simultaneously, screen magnifier users often have to choose or split their attention between the two. Existing screen magnifiers are content-agnostic, so the current way of navigating visualizations is freeform and unguided. The burden of figuring out where to explore while retaining a mental model of the diagram is placed entirely on the user. In this paper, we present Mantis—six prototypes of an automatic, content-aware screen magnification tool designed to aid people who have low vision in the traversal of diagrams. Each design experiments with what sorts of information might be provided to help the user retain a sense of context. Further, they each explore how such a tool might use its knowledge of the diagram’s semantic structure to streamline traversal to and from areas of interest to the user. To this end, we evaluate how these proof-of-concepts improve the user’s navigational experience and reduce the user’s cognitive load.M.Eng
Characterization, Processing, and Synthesis of Extreme-Performance Continuous Carbon Nanotube Network Composites
No full textContinuous carbon nanotube (CNT) networks are an emerging, hierarchically-structured, and commercially available nanomaterial built from countless CNT nanocrystals. These macroscopic yarn materials promise to bridge the gap between microscopic CNT fibers – which are well-known for their superlative material properties – and human-scale fiber reinforcements for extreme-performance composites. Yet because the constituent CNTs interact only via intermolecular forces, network properties fall short of their building blocks. Although these materials show promise as reinforcement in composites, the networks’ low-permeability and tortuous nanoporous structure renders imbibition with liquids like a polymer matrix or surface functionalizing agents challenging. Thus, traditional composite fabrication strategies can be ineffective when applied to CNT yarns, especially commercial products subject to proprietary microstructural manipulation.
Using commercially-available CNT yarns fabricated through floating-catalyst chemical vapor deposition (FCCVD) as model systems, we first explore yarn characteristics which are unique to their hierarchical, bundled-fiber structure, placing focus on the oxygen-rich amorphous carbon phase found in pre-densified, chemically-stretched yarns. A green hydrothermal technique is explored to remove this phase from the surface level inward, allowing for purification and improved infiltrability. However, we find this phase is distinct from previously-reported amorphous carbons found in CNTs, showing it behaves as a matrix which may improve polymer bonding. An analysis of imbibition and fluid transport in these CNT yarns finds that while infiltration of low-viscosity liquids like water is thermodynamically-favored, it is limited when surpassing the threshold of capillary pore percolation. Nevertheless, infiltration in lower-density networks is not only observed, but exploited through the demonstration of dielectric heating in a microwave reactor, where we show fluid imbibed within the network can be boiled to induce swelling and exfoliation of CNT bundles (or conversely, this may be avoided) through optimization of the heating parameters and solvent.
Next, with a firm understanding of the yarn networks’ properties and the impact of various processing effects, we demonstrate two techniques of producing polymer in-situ using dissolved monomers to side-step slow infiltration. The first technique is in-situ interfacial polymerization (ISIP), which is adapted to the yarns studied in this work to yield polyetherimide–CNT yarn composites. When applied to chemically-stretched yarn, specific strengths as high as 2.2 GPa/(g-cm3) are achieved in the flexible and durable yarn composite. We show parameters and conditions which maximize tensile properties and challenges associated with the rapid nature of the process, concluding with the successful demonstration of a roll-to-roll fabrication scheme for producing arbitrary amounts of polymer.
In our second technique, we produce extreme-performance polyimide and polybenzimidazole composites through green in-situ polymerizations (ISSP) in CNTs and macroscopic fiber networks. This approach utilizes superheated water and alcohol as a powerful medium to disperse monomers and initiate polymerization of high-performance coatings within a porous network. We demonstrate ISSP-CNT composites with variable coating morphologies (conformal, shish-kebab, etc.), in-air stability to over 500°C, and doubled specific stiffness and specific strength. Finally, we validate the multifunctional behavior of polyimide-CNT composites by showing a strong, flexible composite can store energy and behave as a free-standing battery electrode.Sc.D
Superconducting Nanowire Integrated Circuits for Scalable Cryogenic Memory
No full textSuperconducting nanowire integrated circuits (SNICs) are a promising class of cryogenic electronics that harness the zero resistance, high kinetic inductance, and nanoscale geometry of ultrathin superconducting wires to implement logic, memory, amplification, and sensing with minimal energy dissipation. Unlike Josephson-junction-based circuits, SNICs support compact, planar layouts compatible with single-layer fabrication and operation in unshielded cryogenic environments. This thesis develops superconducting nanowire memory (SNM) as a scalable implementation of SNICs. A modular cell architecture is introduced, exploiting hysteretic switching and inductive asymmetry to enable nonvolatile digital state storage with zero static power consumption. A hierarchical design framework is established, combining automated layout generation, electrothermal simulation in LTspice, and microscopic modeling using the time-dependent Ginzburg–Landau (TDGL) formalism. To enable scalable integration, this work implements a row–column SNM array layout and demonstrates fabrication across full 4-inch wafers using a planar, singlelayer process. Cryogenic measurements validate reliable operation in both single cells and multi-cell arrays, confirming the predictive accuracy of the design and modeling framework. Tradeoffs in bias current levels, pulse timing, and read/write conditions are systematically evaluated through cryogenic measurements, revealing their impact on bit error rate, operational margins, and energy efficiency across single cells and arrays. Together, these contributions establish SNICs as a viable and extensible platform for cryogenic memory, providing the tools, models, and infrastructure needed to enable broader adoption in quantum computing, neuromorphic systems, and other energy-constrained cryogenic applications.Ph.D