Search

Together with industry experts, we are exploring the potential of head-mounted augmented reality to facilitate safety inspections on high-rise construction sites. A particular concern in the industry is inspecting perimeter safety screens on higher levels of construction sites, intended to prevent falls of people and objects. We aim to support workers performing this inspection task by tracking which parts of the safety screens have been inspected. We use machine learning to automatically detect gaps in the perimeter screens that require closer inspection and remediation and to automate reporting. This work-in-progress paper describes the problem, our early progress, concerns around worker privacy, and the possibilities to mitigate these., Comment: 2 pages, 1 figure, ISMAR24 Workshop Paper

Learning the intents of an agent, defined by its goals or motion style, is often extremely challenging from just a few examples. We refer to this problem as task concept learning and present our approach, Few-Shot Task Learning through Inverse Generative Modeling (FTL-IGM), which learns new task concepts by leveraging invertible neural generative models. The core idea is to pretrain a generative model on a set of basic concepts and their demonstrations. Then, given a few demonstrations of a new concept (such as a new goal or a new action), our method learns the underlying concepts through backpropagation without updating the model weights, thanks to the invertibility of the generative model. We evaluate our method in five domains -- object rearrangement, goal-oriented navigation, motion caption of human actions, autonomous driving, and real-world table-top manipulation. Our experimental results demonstrate that via the pretrained generative model, we successfully learn novel concepts and generate agent plans or motion corresponding to these concepts in (1) unseen environments and (2) in composition with training concepts.

In this work, we explore the limitations of combining models by averaging intermediate features, referred to as model merging, and propose a new direction for achieving collective model intelligence through what we call compatible specialization. Current methods for model merging, such as parameter and feature averaging, struggle to effectively combine specialized models due to representational divergence during fine-tuning. As models specialize to their individual domains, their internal feature representations become increasingly incompatible, leading to poor performance when attempting to merge them for new tasks. We analyze this phenomenon using centered kernel alignment (CKA) and show that as models specialize, the similarity in their feature space structure diminishes, hindering their capacity for collective use. To address these challenges, we investigate routing-based merging strategies, which offer more flexible methods for combining specialized models by dynamically routing across different layers. This allows us to improve on existing methods by combining features from multiple layers rather than relying on fixed, layer-wise combinations. However, we find that these approaches still face limitations when layers within models are representationally incompatible. Our findings highlight the importance of designing new approaches for model merging that operate on well-defined input and output spaces, similar to how humans communicate through language rather than intermediate neural activations.

The correct specification of reward models is a well-known challenge in reinforcement learning. Hand-crafted reward functions often lead to inefficient or suboptimal policies and may not be aligned with user values. Reinforcement learning from human feedback is a successful technique that can mitigate such issues, however, the collection of human feedback can be laborious. Recent works have solicited feedback from pre-trained large language models rather than humans to reduce or eliminate human effort, however, these approaches yield poor performance in the presence of hallucination and other errors. This paper studies the advantages and limitations of reinforcement learning from large language model feedback and proposes a simple yet effective method for soliciting and applying feedback as a potential-based shaping function. We theoretically show that inconsistent rankings, which approximate ranking errors, lead to uninformative rewards with our approach. Our method empirically improves convergence speed and policy returns over commonly used baselines even with significant ranking errors, and eliminates the need for complex post-processing of reward functions., Comment: 13 pages, 8 figures, The 2024 Conference on Empirical Methods in Natural Language Processing

Multi-Agent Reinforcement Learning (MARL) methods have shown promise in enabling agents to learn a shared communication protocol from scratch and accomplish challenging team tasks. However, the learned language is usually not interpretable to humans or other agents not co-trained together, limiting its applicability in ad-hoc teamwork scenarios. In this work, we propose a novel computational pipeline that aligns the communication space between MARL agents with an embedding space of human natural language by grounding agent communications on synthetic data generated by embodied Large Language Models (LLMs) in interactive teamwork scenarios. Our results demonstrate that introducing language grounding not only maintains task performance but also accelerates the emergence of communication. Furthermore, the learned communication protocols exhibit zero-shot generalization capabilities in ad-hoc teamwork scenarios with unseen teammates and novel task states. This work presents a significant step toward enabling effective communication and collaboration between artificial agents and humans in real-world teamwork settings., Comment: Accepted to Neurips 2024, 19 pages, 10 figures

This study provides an analysis of atmospheric neutrino oscillations at the ESSnuSB far detector facility. The prospects of the two cylindrical Water Cherenkov detectors with a total fiducial mass of 540 kt are investigated over 10 years of data taking in the standard three-flavor oscillation scenario. We present the confidence intervals for the determination of mass ordering, $\theta_{23}$ octant as well as for the precisions on $\sin^2\theta_{23}$ and $|\Delta m_{31}^2|$. It is shown that mass ordering can be resolved by $3\sigma$ CL ($5\sigma$ CL) after 4 years (10 years) regardless of the true neutrino mass ordering. Correspondingly, the wrong $\theta_{23}$ octant could be excluded by $3\sigma$ CL after 4 years (8 years) in the case where the true neutrino mass ordering is normal ordering (inverted ordering). The results presented in this work are complementary to the accelerator neutrino program in the ESSnuSB project., Comment: 14 pages, 7 figures and 2 tables, accepted for publication in the Journal of High Energy Physics

The Super-Kamiokande and T2K collaborations present a joint measurement of neutrino oscillation parameters from their atmospheric and beam neutrino data. It uses a common interaction model for events overlapping in neutrino energy and correlated detector systematic uncertainties between the two datasets, which are found to be compatible. Using 3244.4 days of atmospheric data and a beam exposure of $19.7(16.3) \times 10^{20}$ protons on target in (anti)neutrino mode, the analysis finds a 1.9$\sigma$ exclusion of CP-conservation (defined as $J_{CP}=0$) and a preference for the normal mass ordering., Comment: 12 pages, 4 figures

We propose a computationally efficient and systematically convergent approach for elastodynamics simulations. We recast the second-order dynamical equation of elastodynamics into an equivalent first-order system of coupled equations, so as to express the solution in the form of a Magnus expansion. With any spatial discretization, it entails computing the exponential of a matrix acting upon a vector. We employ an adaptive Krylov subspace approach to inexpensively and and accurately evaluate the action of the exponential matrix on a vector. In particular, we use an apriori error estimate to predict the optimal Kyrlov subspace size required for each time-step size. We show that the Magnus expansion truncated after its first term provides quadratic and superquadratic convergence in the time-step for nonlinear and linear elastodynamics, respectively. We demonstrate the accuracy and efficiency of the proposed method for one linear (linear cantilever beam) and three nonlinear (nonlinear cantilever beam, soft tissue elastomer, and hyperelastic rubber) benchmark systems. For a desired accuracy in energy, displacement, and velocity, our method allows for $10-100\times$ larger time-steps than conventional time-marching schemes such as Newmark-$\beta$ method. Computationally, it translates to a $\sim$$1000\times$ and $\sim$$10-100\times$ speed-up over conventional time-marching schemes for linear and nonlinear elastodynamics, respectively.

Neutrino oscillation experiments provide a unique window in exploring several new physics scenarios beyond the standard three flavour. One such scenario is quantum decoherence in neutrino oscillation which tends to destroy the interference pattern of neutrinos reaching the far detector from the source. In this work, we study the decoherence in neutrino oscillation in the context of the ESSnuSB experiment. We consider the energy-independent decoherence parameter and derive the analytical expressions for P$_{\mu e}$ and P$_{\mu \mu}$ probabilities in vacuum. We have computed the capability of ESSnuSB to put bounds on the decoherence parameters namely, $\Gamma_{21}$ and $\Gamma_{32}$ and found that the constraints on $\Gamma_{21}$ are competitive compared to the DUNE bounds and better than the most stringent LBL ones from MINOS/MINOS+. We have also investigated the impact of decoherence on the ESSnuSB measurement of the Dirac CP phase $\delta_{\rm CP}$ and concluded that it remains robust in the presence of new physics., Comment: 30 pages, 9 figures, 2 tables, Version accepted for publication in JHEP

Enhancing simulation environments to replicate real-world driver behavior is essential for developing Autonomous Vehicle technology. While some previous works have studied the yielding reaction of lag vehicles in response to a merging car at highway on-ramps, the possible lane-change reaction of the lag car has not been widely studied. In this work we aim to improve the simulation of the highway merge scenario by including the lane-change reaction in addition to yielding behavior of main-lane lag vehicles, and we evaluate two different models for their ability to capture this reactive lane-change behavior. To tune the payoff functions of these models, a novel naturalistic dataset was collected on U.S. highways that provided several hours of merge-specific data to learn the lane change behavior of U.S. drivers. To make sure that we are collecting a representative set of different U.S. highway geometries in our data, we surveyed 50,000 U.S. highway on-ramps and then selected eight representative sites. The data were collected using roadside-mounted lidar sensors to capture various merge driver interactions. The models were demonstrated to be configurable for both keep-straight and lane-change behavior. The models were finally integrated into a high-fidelity simulation environment and confirmed to have adequate computation time efficiency for use in large-scale simulations to support autonomous vehicle development., Comment: 10 pages, 7 figures, submitted to IEEE Intelligent Vehicles Symposium (IV) 2024

Recent advances in visual reasoning (VR), particularly with the aid of Large Vision-Language Models (VLMs), show promise but require access to large-scale datasets and face challenges such as high computational costs and limited generalization capabilities. Compositional visual reasoning approaches have emerged as effective strategies; however, they heavily rely on the commonsense knowledge encoded in Large Language Models (LLMs) to perform planning, reasoning, or both, without considering the effect of their decisions on the visual reasoning process, which can lead to errors or failed procedures. To address these challenges, we introduce HYDRA, a multi-stage dynamic compositional visual reasoning framework designed for reliable and incrementally progressive general reasoning. HYDRA integrates three essential modules: a planner, a Reinforcement Learning (RL) agent serving as a cognitive controller, and a reasoner. The planner and reasoner modules utilize an LLM to generate instruction samples and executable code from the selected instruction, respectively, while the RL agent dynamically interacts with these modules, making high-level decisions on selection of the best instruction sample given information from the historical state stored through a feedback loop. This adaptable design enables HYDRA to adjust its actions based on previous feedback received during the reasoning process, leading to more reliable reasoning outputs and ultimately enhancing its overall effectiveness. Our framework demonstrates state-of-the-art performance in various VR tasks on four different widely-used datasets., Comment: Accepted by ECCV2024. Project page: https://hydra-vl4ai.github.io/

The high-luminosity upgrade of the LHC brings unprecedented requirements for real-time and precision bunch-by-bunch online luminosity measurement and beam-induced background monitoring. A key component of the CMS Beam Radiation, Instrumentation and Luminosity system is a stand-alone luminometer, the Fast Beam Condition Monitor (FBCM), which is fully independent from the CMS central trigger and data acquisition services and able to operate at all times with a triggerless readout. FBCM utilizes a dedicated front-end application-specific integrated circuit (ASIC) to amplify the signals from CO$_2$-cooled silicon-pad sensors with a timing resolution of a few nanoseconds, which enables the measurement of the beam-induced background. FBCM uses a modular design with two half-disks of twelve modules at each end of CMS, with four service modules placed close to the outer edge to reduce radiation-induced aging. The electronics system design adapts several components from the CMS Tracker for power, control and read-out functionalities. The dedicated FBCM23 ASIC contains six channels and adjustable shaping time to optimize the noise with regards to sensor leakage current. Each ASIC channel outputs a single binary high-speed asynchronous signal carrying time-of-arrival and time-over-threshold information. The chip output signal is digitized, encoded and sent via a radiation-hard gigabit transceiver and an optical link to the back-end electronics for analysis. This paper reports on the updated design of the FBCM detector and the ongoing testing program., Comment: 16th Topical Seminar on Innovative Particle and Radiation Detectors (IPRD23) 2023 Sept 25-29 Siena, Italy

Topological semimetal antiferromagnets provide a rich source of exotic topological states which can be controlled by manipulating the orientation of the N\'eel vector, or by modulating the lattice parameters through strain. We investigate via ${ab\ initio}$ density functional theory calculations, the effects of shear strain on the bulk and surface states n two antiferromagnetic EuCd$_2$As$_2$ phases with out-of-plane and in-plane spin configurations. When magnetic moments are along the $\textit{c}$-axis, a $3\%$ longitudinal or diagonal shear strain can tune the Dirac semimetal phase to an axion insulator phase, characterized by the parity-based invariant $\eta_{4I} = 2$. For an in-plane magnetic order, the axion insulator phase remains robust under all shear strains. We further find that for both magnetic orders, the bulk gap increases and a surface gap opens on the (001) surface up to 16 meV. Because of a nonzero $\eta_{4I}$ index and gapped states on the (001) surface, hinge modes are expected to happen on the side surface states between those gapped surface states. This result can provide a valuable insight in the realization of the long-sought axion states., Comment: 5 pages, 4 figures

In this paper we study non-standard interactions mediated by a scalar field (SNSI) in the context of ESSnuSB experiment. In particular we study the capability of ESSnuSB to put bounds on the SNSI parameters and also study the impact of SNSI in the measurement of the leptonic CP phase $\delta_{\rm CP}$. Existence of SNSI modifies the neutrino mass matrix and this modification can be expressed in terms of three diagonal real parameters ($\eta_{ee}$, $\eta_{\mu\mu}$ and $\eta_{\tau\tau}$) and three off-diagonal complex parameters ($\eta_{e \mu}$, $\eta_{e\tau}$ and $\eta_{\mu\tau}$). Our study shows that the upper bounds on the parameters $\eta_{\mu\mu}$, $\eta_{\tau\tau}$ and $\eta_{\mu\tau}$ depend upon how $\Delta m^2_{31}$ is minimized in the theory. However, this is not the case when one tries to measure the impact of SNSI on $\delta_{\rm CP}$. Further, we show that the CP sensitivity of ESSnuSB can be completely lost for certain values of $\eta_{ee}$ and $\eta_{\mu\tau}$ for which the appearance channel probability becomes independent of $\delta_{\rm CP}$., Comment: 14 pages, 6 figures, 2 tables, Version accepted for publication in Phys. Rev. D

Merging into dense highway traffic for an autonomous vehicle is a complex decision-making task, wherein the vehicle must identify a potential gap and coordinate with surrounding human drivers, each of whom may exhibit diverse driving behaviors. Many existing methods consider other drivers to be dynamic obstacles and, as a result, are incapable of capturing the full intent of the human drivers via this passive planning. In this paper, we propose a novel dual control framework based on Model Predictive Path-Integral control to generate interactive trajectories. This framework incorporates a Bayesian inference approach to actively learn the agents' parameters, i.e., other drivers' model parameters. The proposed framework employs a sampling-based approach that is suitable for real-time implementation through the utilization of GPUs. We illustrate the effectiveness of our proposed methodology through comprehensive numerical simulations conducted in both high and low-fidelity simulation scenarios focusing on autonomous on-ramp merging., Comment: 7 pages, 3 figures

In this paper, we develop a control framework for the coordination of multiple robots as they navigate through crowded environments. Our framework comprises of a local model predictive control (MPC) for each robot and a social long short-term memory model that forecasts pedestrians' trajectories. We formulate the local MPC formulation for each individual robot that includes both individual and shared objectives, in which the latter encourages the emergence of coordination among robots. Next, we consider the multi-robot navigation and human-robot interaction, respectively, as a potential game and a two-player game, then employ an iterative best response approach to solve the resulting optimization problems in a centralized and distributed fashion. Finally, we demonstrate the effectiveness of coordination among robots in simulated crowd navigation.

Crowd navigation has received increasing attention from researchers over the last few decades, resulting in the emergence of numerous approaches aimed at addressing this problem to date. Our proposed approach couples agent motion prediction and planning to avoid the freezing robot problem while simultaneously capturing multi-agent social interactions by utilizing a state-of-the-art trajectory prediction model i.e., social long short-term memory model (Social-LSTM). Leveraging the output of Social-LSTM for the prediction of future trajectories of pedestrians at each time-step given the robot's possible actions, our framework computes the optimal control action using Model Predictive Control (MPC) for the robot to navigate among pedestrians. We demonstrate the effectiveness of our proposed approach in multiple scenarios of simulated crowd navigation and compare it against several state-of-the-art reinforcement learning-based methods., Comment: 7 pages, 3 figures, 6 tables

We report an updated measurement of the $\nu_{\mu}$-induced, and the first measurement of the $\bar{\nu}_{\mu}$-induced coherent charged pion production cross section on $^{12}C$ nuclei in the T2K experiment. This is measured in a restricted region of the final-state phase space for which $p_{\mu,\pi} > 0.2$ GeV, $\cos(\theta_{\mu}) > 0.8$ and $\cos(\theta_{\pi}) > 0.6$, and at a mean (anti)neutrino energy of 0.85 GeV using the T2K near detector. The measured $\nu_{\mu}$ CC coherent pion production flux-averaged cross section on $^{12}C$ is $(2.98 \pm 0.37 (stat.) \pm 0.31 (syst.) \substack{ +0.49 \\ -0.00 } \mathrm{ (Q^2\,model)}) \times 10^{-40}~\mathrm{cm}^{2}$. The new measurement of the $\bar{\nu}_{\mu}$-induced cross section on $^{12}{C}$ is $(3.05 \pm 0.71 (stat.) \pm 0.39 (syst.) \substack{ +0.74 \\ -0.00 } \mathrm{(Q^2\,model)}) \times 10^{-40}~\mathrm{cm}^{2}$. The results are compatible with both the NEUT 5.4.0 Berger-Sehgal (2009) and GENIE 2.8.0 Rein-Sehgal (2007) model predictions.

The ENUBET project is aimed at designing and experimentally demonstrating the concept of monitored neutrino beams. These novel beams are enhanced by an instrumented decay tunnel, whose detectors reconstruct large-angle charged leptons produced in the tunnel and give a direct estimate of the neutrino flux at the source. These facilities are thus the ideal tool for high-precision neutrino cross-section measurements at the GeV scale because they offer superior control of beam systematics with respect to existing facilities. In this paper, we present the first end-to-end design of a monitored neutrino beam capable of monitoring lepton production at the single particle level. This goal is achieved by a new focusing system without magnetic horns, a 20 m normal-conducting transfer line for charge and momentum selection, and a 40 m tunnel instrumented with cost-effective particle detectors. Employing such a design, we show that percent precision in cross-section measurements can be achieved at the CERN SPS complex with existing neutrino detectors., Comment: 24 pages, 33 figures

This paper presents a comparative analysis of three distinct methods used to calculate the collisional energy loss of heavy quarks in Quark-Gluon Plasma. The study focuses on the calculation of the nuclear suppression factor of charm quarks in Pb-Pb collisions at $\sqrt{S_{NN}} = 5.02$ TeV. All three models are examined using the same numerical evolution based on the well-known Fokker-Planck equation by considering critical phenomena like a non-equilibrium state at the onset of heavy ion collision. The outcomes of each approach are compared with the latest data from ALICE and ATLAS experiments spanning from 2018 to 2022. This study aims to compare the degree of agreement between each approach and recently obtained experimental data, in the intermediate and high $P_T$ regions., Comment: 23 Pages, 7 Figures, 4 Tables

In this article we propose a scheme for solving the coupled time-fractional nonlocal diffusion problem. The scheme consist of fractional Crank-Nicolson method with Galerkin finite element method (FEM) and Newton's method. We derive \emph{a priori} error estimates for fully-discrete solutions in $L^2$ and $H^1_0$ norms. Results based on the usual finite element method are provided to confirm the theoretical estimates.

In this article, we propose numerical scheme for solving a multi-term time-fractional nonlocal parabolic partial differential equation (PDE). The scheme comprises $L2$-$1_{\sigma}$ scheme on a graded mesh in time and Galerkin finite element method (FEM) in space. We present the discrete fractional Gr$\ddot{{o}}$nwall inequality for $L2$-$1_{\sigma}$ scheme in case of multi-term time-fractional derivative, which is a multi-term analogue of~\cite[Lemma 4.1]{[r16]}. We derive \textit{a priori} bound and error estimate for the fully-discrete solution. The theoretical results are confirmed via numerical experiments. We should note that, though the way of proving the discrete fractional Gr$\ddot{{o}}$nwall inequality is similar to~\cite{[r5]}, the calculation parts are more complicated in this article., Comment: arXiv admin note: text overlap with arXiv:2306.02576

In this work, a time-fractional nonlocal diffusion equation is considered. Based on the $L2$-$1_{\sigma}$ scheme on a graded mesh in time and the standard finite element method (FEM) in space, the fully-discrete $L2$-$1_{\sigma}$ finite element method is investigated for a time-fractional nonlocal diffusion problem. We prove the existence and uniqueness of fully-discrete solution. The $\alpha$-robust error bounds are derived, i.e. bounds remain valid as $\alpha$ $\rightarrow {1}^{-},$ where $\alpha \ \in (0,1)$ is the order of a temporal fractional derivative. The numerical experiments are presented to justify the theoretical findings.

Three challenges limit the progress of robot learning research: robots are expensive (few labs can participate), everyone uses different robots (findings do not generalize across labs), and we lack internet-scale robotics data. We take on these challenges via a new benchmark: Train Offline, Test Online (TOTO). TOTO provides remote users with access to shared robotic hardware for evaluating methods on common tasks and an open-source dataset of these tasks for offline training. Its manipulation task suite requires challenging generalization to unseen objects, positions, and lighting. We present initial results on TOTO comparing five pretrained visual representations and four offline policy learning baselines, remotely contributed by five institutions. The real promise of TOTO, however, lies in the future: we release the benchmark for additional submissions from any user, enabling easy, direct comparison to several methods without the need to obtain hardware or collect data., Comment: Accepted to ICRA 2023

Efficient compression of correlated data is essential to minimize communication overload in multi-sensor networks. In such networks, each sensor independently compresses the data and transmits them to a central node due to limited communication bandwidth. A decoder at the central node decompresses and passes the data to a pre-trained machine learning-based task to generate the final output. Thus, it is important to compress the features that are relevant to the task. Additionally, the final performance depends heavily on the total available bandwidth. In practice, it is common to encounter varying availability in bandwidth, and higher bandwidth results in better performance of the task. We design a novel distributed compression framework composed of independent encoders and a joint decoder, which we call neural distributed principal component analysis (NDPCA). NDPCA flexibly compresses data from multiple sources to any available bandwidth with a single model, reducing computing and storage overhead. NDPCA achieves this by learning low-rank task representations and efficiently distributing bandwidth among sensors, thus providing a graceful trade-off between performance and bandwidth. Experiments show that NDPCA improves the success rate of multi-view robotic arm manipulation by 9% and the accuracy of object detection tasks on satellite imagery by 14% compared to an autoencoder with uniform bandwidth allocation.