Search

This paper investigates runtime monitoring of perception systems. Perception is a critical component of high-integrity applications of robotics and autonomous systems, such as self-driving cars. In these applications, failure of perception systems may put human life at risk, and a broad adoption of these technologies requires the development of methodologies to guarantee and monitor safe operation. Despite the paramount importance of perception, currently there is no formal approach for system-level perception monitoring. In this paper, we formalize the problem of runtime fault detection and identification in perception systems and present a framework to model diagnostic information using a diagnostic graph. We then provide a set of deterministic, probabilistic, and learning-based algorithms that use diagnostic graphs to perform fault detection and identification. Moreover, we investigate fundamental limits and provide deterministic and probabilistic guarantees on the fault detection and identification results. We conclude the paper with an extensive experimental evaluation, which recreates several realistic failure modes in the LGSVL open-source autonomous driving simulator, and applies the proposed system monitors to a state-of-the-art autonomous driving software stack (Baidu's Apollo Auto). The results show that the proposed system monitors outperform baselines, have the potential of preventing accidents in realistic autonomous driving scenarios, and incur a negligible computational overhead.

The neutron diffraction and nuclear magnetic resonance (NMR) measurements have been used to microscopically analyze the magnetic structure in the $S = 1/2$ kagome antiferromagnet CdCu$_{3}$(OH)$_{6}$(NO$_{3}$)$_{2}\cdot $H$_{2}$O. Below the magnetic ordering temperature $T_N\simeq 4$ K, magnetic Bragg reflections at (110) and (100) were found in the neutron diffraction pattern, which suggests a $q=0$ magnetic structure. Furthermore, the vector spin chirality for the $q=0$ structure was successfully identified from the internal field direction obtained by the $^{14}$N-NMR measurement. Our findings point to a chirality-ordered magnetic structure with negative vector chirality and $\langle 100 \rangle$ anisotropy., Comment: 6 pages, 5 figures

The principal Hugoniot of carbon, initially diamond, was measured from 3 to 80 TPa (30 to 800 million atmospheres), the highest pressure ever achieved, using radiography of spherically-converging shocks. The shocks were generated by ablation of a plastic coating by soft x-rays in a laser-heated hohlraum at the National Ignition Facility (NIF). Experiments were performed with low and high drive powers, spanning different but overlapping pressure ranges. The radius-time history of the shock, and the profile of mass density behind, were determined by profile-matching from a time-resolved x-ray radiograph across the diameter of the sphere. Above ~50 TPa, the heating induced by the shock was great enough to ionize a significant fraction of K-shell electrons, reducing the opacity to the 10.2 keV probe x-rays. The opacity and mass density were deduced simultaneously using the constraint that the total mass of the sample was constant. The Hugoniot and opacity were consistent with density functional theory calculations of the electronic states and equation of state (EOS), and varied significantly from theoretical Hugoniots based on Thomas-Fermi theory. Theoretical models used to predict the compressibility of diamond ablator experiments at the NIF, producing the highest neutron yields so far from inertial confinement fusion experiments, are qualitatively consistent with our EOS measurements but appear to overpredict the compressibility slightly. These measurements help to evaluate theoretical techniques and constrain wide-range EOS models applicable to white dwarf stars, which are the ultimate evolutionary form of at least 97% of stars in the galaxy., Comment: arXiv admin note: text overlap with arXiv:2203.08891

We previously reported an experimental platform to induce a spherically-convergent shock in a sample using laser-driven ablation, probed with time-resolved x-ray radiography, and an analysis method to deduce states along the principal shock Hugoniot simultaneously with the x-ray opacity. We have now developed a modified method of analysis that is numerically better-conditioned and faster, and usually provides a better representation of the radiograph with correspondingly lower uncertainties. The previous approach was based on optimizing parameters in a model of the density distribution as a function of radius and time, warped to follow loci such as the shock and the outside of the sample. The converging shock configuration can be described more efficiently in terms of the shocked density and sound speed, expressed as functions of the shock speed Studies of the Hugoniot from various theoretical equations of state (EOS) indicate that, in the typical range of states explored by these experiments, these functions can be described by low-order polynomials. Similarly, few-parameter functions were found suitable for representing the variation of x-ray opacity with shock pressure. This approach was found to perform better in most cases than an alternative method based on parameterization of the EOS.

The quantum approximate optimization algorithm/quantum alternating operator ansatz (QAOA) is a heuristic to find approximate solutions of combinatorial optimization problems. Most literature is limited to quadratic problems without constraints. However, many practically relevant optimization problems do have (hard) constraints that need to be fulfilled. In this article, we present a framework for constructing mixing operators that restrict the evolution to a subspace of the full Hilbert space given by these constraints; We generalize the "XY"-mixer designed to preserve the subspace of "one-hot" states to the general case of subspaces given by a number of computational basis states. We expose the underlying mathematical structure which reveals more of how mixers work and how one can minimize their cost in terms of number of CX gates, particularly when Trotterization is taken into account. Our analysis also leads to valid Trotterizations for "XY"-mixer with fewer CX gates than is known to date. In view of practical implementations, we also describe algorithms for efficient decomposition into basis gates. Several examples of more general cases are presented and analyzed.

We present a randomized algorithm that computes single-source shortest paths (SSSP) in $O(m\log^8(n)\log W)$ time when edge weights are integral and can be negative. This essentially resolves the classic negative-weight SSSP problem. The previous bounds are $\tilde O((m+n^{1.5})\log W)$ [BLNPSSSW FOCS'20] and $m^{4/3+o(1)}\log W$ [AMV FOCS'20]. Near-linear time algorithms were known previously only for the special case of planar directed graphs [Fakcharoenphol and Rao FOCS'01]. In contrast to all recent developments that rely on sophisticated continuous optimization methods and dynamic algorithms, our algorithm is simple: it requires only a simple graph decomposition and elementary combinatorial tools. In fact, ours is the first combinatorial algorithm for negative-weight SSSP to break through the classic $\tilde O(m\sqrt{n}\log W)$ bound from over three decades ago [Gabow and Tarjan SICOMP'89]., Comment: Simplified algorithms and minor corrections throughout the paper

Stroke rehabilitation seeks to increase neuroplasticity through the repeated practice of functional motions, but may have minimal impact on recovery because of insufficient repetitions. The optimal training content and quantity are currently unknown because no practical tools exist to measure them. Here, we present PrimSeq, a pipeline to classify and count functional motions trained in stroke rehabilitation. Our approach integrates wearable sensors to capture upper-body motion, a deep learning model to predict motion sequences, and an algorithm to tally motions. The trained model accurately decomposes rehabilitation activities into component functional motions, outperforming competitive machine learning methods. PrimSeq furthermore quantifies these motions at a fraction of the time and labor costs of human experts. We demonstrate the capabilities of PrimSeq in previously unseen stroke patients with a range of upper extremity motor impairment. We expect that these advances will support the rigorous measurement required for quantitative dosing trials in stroke rehabilitation.

Over the course of six years, a diverse group of school districts and charter management organizations in California ran a substantial experiment in implementing the Next Generation Science Standards (NGSS) in elementary and middle grades. WestEd led an extensive evaluation of that effort, the California K-8 NGSS Early Implementers Initiative, resulting in more than a dozen evaluation reports highlighting important themes and learnings. All reports are intended to be helpful to administrators at the school and district levels, education policymakers, and people charged with designing and/or delivering science professional learning. This report on highlights from the Early Implementers Initiative is the final report of the series. It offers high-level, major lessons learned, challenges, and strategies gleaned from the evaluation series. It also offers recommendations for policymakers and administrators for implementing the Next Generation Science Standards in their own districts.

This 13th report in WestEd's evaluation of the K-8 Early Implementers Initiative for the Next Generation Science Standards (NGSS) provides an extensive response to the following question: What does NGSS science teaching look like in the classroom? The report first discusses Initiative teachers' experiences with four instructional practices that promote equitable instruction, an explicit feature of the NGSS. The report then illustrates how Early Implementer teachers enacted three additional features of NGSS instruction: (1) integrating the three dimensions of the NGSS; (2) using real-world scientific phenomena and/or engineering problems to launch and drive instructional lessons and units; and (3) having students do investigations in ways that give them the responsibility and opportunity for learning. Each vignette is followed by commentary explaining how key features of the NGSS were incorporated into the lesson, with the intention of describing the NGSS teaching in enough detail to help administrators and policymakers recognize when they are seeing NGSS-based instruction and when they aren't. The report also briefly relates specific ways that Initiative teachers have advanced in their NGSS science teaching over the years of the Initiative, and how the Initiative prepared teachers for such teaching. The report concludes with recommendations for how administrators can support teachers in implementing NGSS teaching.

The NGSS Early Implementers Initiative was created to help eight California school districts and two charter management organizations, supported by WestEd's K-12 Alliance, implement the Next Generation Science Standards (NGSS). Many educational initiatives are funded for only a couple of years. This unique initiative spanned an extraordinary six years, during which eight school districts worked toward districtwide implementation of the Next Generation Science Standards (NGSS), which call for teachers to transform their instructional practice. This 12th report in WestEd's evaluation series for policymakers, school and district administrators, and professional learning specialists describes the Initiative's scale-up in its later years to reach all K-8 science teachers. Specifically, the report addresses the following questions: (1) In contrast to focusing in Years 1-4 on developing Teacher Leaders, what strategies did districts use in Years 5-6 to reach all other K-8 teachers of science (called "expansion teachers" in this report)?; (2) What impacts has the Initiative had on expansion teachers?; (3) Which professional learning strategies have been most and least effective for influencing the practice of expansion teachers?; and (4) What special attention was paid to providing administrators with professional learning to prompt their support of NGSS implementation?

The California Next Generation Science Standards (CA NGSS) Toolkit for Instructional Materials Evaluation (TIME) is a suite of tools and processes for evaluating and selecting instructional materials aligned with the CA NGSS. This 11th report in the NGSS Early Implementers Initiative evaluation series is intended for school and district administrators, leaders of science professional learning, and state policymakers. It provides an overview of the full TIME process, including participants' perceptions, a detailed description of the statewide TIME trainings of 2018-19, and a vignette that illustrates a portion of the TIME process.