StereoBit is an embedded application for the SpaceCube family of hybrid onboard processors developed by NASA's Goddard Space Flight Center under sponsorship of the Earth Science Technology Office. StereoBit applies structure from motion algorithms to measure disparities between clouds being tracked from multiple angles to enable 3D resolution of atmospheric motion, which is an important Decadal Survey objective. Our Advanced Information Systems Technology project has the objectives of developing and demonstrating StereoBit for a future mission architecture employing a constellation of CubeSats and to mature the capabilities of the community to program the SpaceCube- family of processors for science applications.

Imitation learning has proven to be useful for many real-world problems, but approaches such as behavioral cloning suffer from data mismatch and compounding error issues. One attempt to address these limitations is the DAgger algorithm, which uses the state distribution induced by the novice to sample corrective actions from the expert. Such sampling schemes, however, require the expert to provide action labels without being fully in control of the system. This can decrease safety and, when using humans as experts, is likely to degrade the quality of the collected labels due to perceived actuator lag. In this work, we propose HG-DAgger, a variant of DAgger that is more suitable for interactive imitation learning from human experts in real-world systems. In addition to training a novice policy, HG-DAgger also learns a safety threshold for a model-uncertainty-based risk metric that can be used to predict the performance of the fully trained novice in different regions of the state space. We evaluate our method on both a simulated and real-world autonomous driving task, and demonstrate improved performance over both DAgger and behavioral cloning.

Embedded systems are increasingly required to handle code of various qualities that must often be isolated, yet predictably share resources. This has motivated the isolation of, for example, mission-critical code from best-effort features using isolation structures such as virtualization. Such systems usually focus on limiting interference between subsystems, which complicates the increasingly common functional dependencies between them. Though isolation must be paramount, the fundamental goal of efficiently sharing hardware motivates a principled mechanism for cooperating between subsystems. This paper introduces Temporal Capabilities (TCaps) which integrate CPU management into a capability-based access-control system and distribute authority for scheduling. In doing so, the controlled temporal coordination between subsystems becomes a first-class concern of the system. By enabling temporal delegations to accompany activations and requests for service, we apply TCaps to a virtualization environment with a shared VM for orchestrating I/O. We show that TCaps, unlike prioritizations and carefully chosen budgets, both meet deadlines for a hard real-time subsystem, and maintain high throughput for a best-effort subsystem.

Static textual analysis techniques have been recently applied to process and synthesize source code. The underlying tenet is that important information is embedded in code identifiers and internal code comments. Such information can be analyzed to provide automatic aid for several software engineering activities. To facilitate this line of work, we present STAC, a tool for supporting Static Textual Analysis of Code. STAC is designed as a light-weight stand-alone tool that provides a practical one-stop solution for code indexing. Code indexing is the process of extracting important textual information from source code. Accurate indexing has been found to significantly influence the performance of code retrieval and analysis methods. STAC provides features for extracting and processing textual patterns found in Java, C++, and C# code artifacts. These features include identifier splitting, stemming, lemmatization, and spell-checking. STAC is also provided as an API to help researchers to integrate basic code indexing features into their code.

Model update is an approach to enhance model checking functions by providing computer aided modifications in system development [2, 9]. It has been observed that one major obstacle restricting the application of this approach, e.g. CTL model update [15], is that the update has to take the entire system model into account, and that is usually not feasible for large scale domains. In this paper, we develop a tree-like local model update approach under the framework of ACTL – a widely used fragment of CTL in property specification. We define a bisimulation based minimal change principle on tree-like local model update, reveal its relationship to traditional belief update and provide essential semantic characterizations. We also investigate primary semantic and computational properties in relation to tree-like local model update. Finally we briefly describe the update system prototype that we have implemented and summarize our experimental results.

Research in the field of biometrics depends on the effective management of large amounts of data and computation. Current research projects in biometrics acquire many terabytes of images and video of subjects in many different modes and situations, annotated with detailed metadata. To study the effectiveness of new algorithms for identifying people, researchers must exhaustively compare large numbers of measurements with a variety of custom functions. The quality of the end results is often dependent upon the sheer amount of data marshalled to support it.To address these challenges, we are constructing BXGrid, an end-to-end computing system for conducting biometrics research. BXGrid assists with the entire research process from data acquisition all the way to generating results for publication. Because the entire chain of research is kept consistently within one system, multiple users may easily share tools and results, building off of each other's work. BXGrid also helps to ensure scientific integrity by automating a variety of consistency checks, external data audits, and reproduction of existing results.

Classroom note taking tends to be a private activity, hiding a wealth of knowledge in both content and method. With the advent of the web, whose technology and culture seemingly invites everyone to share everything, we are prompted to ask how making note taking a public activity -- noteblogging -- might advance learning. What does a blog about a computer science classroom look like? What supports are needed to enable noteblogging as a form of auxiliary instruction? In this paper we overview the design and use of noteblogging as part of the Ubiquitous Presenter digital classroom system. With NoteBlogger, students with Tablet PCs take handwritten notes digitally on top of the instructor's slides, and their notes are instantly reviewable by other students in class using a web browser. We examine the impact of noteblogging on a CS1 course through interviews with bloggers (to reveal their motivation and understand their choice of content) and blog watchers (to learn when and why they watch). We also analyze the blogs in this CS1 course to identify their educational contribution. We find that noteblogging enables a unique classroom participation model that specifically engages more advanced students. Blog content spans many levels of intellectual engagement, which can support a range of learners in CS1 as well as perhaps model for them various levels of reflection.

Peer review is widely recognized for advancing student learning, in particular for developing reflective processes like critical thinking. The classroom is ripe for peer review because the subject matter is fresh and in-depth interactivity is possible. Yet the limited time available in class conflicts with peer review's deliberative nature. We hypothesize that peer review -- at least the initial stages of it -- can be supported in the classroom with tools for facilitating the rapid identification of interesting issues for discussion. The potential benefits of such a tool include: furthering the student-focus of in-class active learning activities, further implanting critical analysis skills through frequent in-class use, supporting immediate feedback, and enabling comparison of student and instructor-modeled critical analysis.This paper explores tool support for in-class lightweight preliminary peer-review (LPPR): peer review that is instigated in the classroom, but does not necessarily end there. We proposed that students classify peer solutions in 4 dimensions: correctness, comprehension (e.g., "do I understand this solution"), worthiness for discussion, and similarity to the evaluator's own solution. We designed an LPPR extension to Ubiquitous Presenter, and then conducted an exploratory study in a mock classroom setting. We found that LPPR can quickly identify a subset of student solutions that warrant immediate discussion, and that modest amounts of reflection arise from the LPPR process.

Research has shown that educational technology can broaden and enhance the use of active learning in large classrooms. An educational technology platform often relies on students to bring laptops or specialized wireless devices like clickers to interact through the system. Mobile phones are an attractive alternative, as most students already possess them, they have more capabilities than dedicated clickers, and yet are small enough to minimize interference with note taking on a classroom desk.This paper presents the design and use of a mobile phone extension to Ubiquitous Presenter, which allows students to submit solutions to active learning exercises in the form of text or photo messages. In an exploratory study, students found that text messaging worked well for exercises with multiple choice or short answers. Entering symbols common to computer science was difficult. Many problems were more suitable to photo messaging of a handwritten answer, although image quality must be managed. The phone's small size left space for the use of a notebook. The students had concerns about the message charges that would accrue in use. In conclusion, we offer recommendations to instructors and system designers interested in leveraging mobile phones to increase communication in the classroom.