arXiv subjects

Gang Chen

Publications and source records attributed to Gang Chen.

At least 163 records · Page 9Linked to original sources

arXiv2025

Pushing Through Clutter With Movability Awareness of Blocking Obstacles

Navigation Among Movable Obstacles (NAMO) poses a challenge for traditional path-planning methods when obstacles block the path, requiring push actions to reach the goal. We propose a framework that enables movability-aware planning to overcome this challenge without relying on explicit obstacle placement. Our framework integrates a global Semantic Visibility Graph and a local Model Predictive Path Integral (SVG-MPPI) approach to efficiently sample rollouts, taking into account the continuous range of obstacle movability. A physics engine is adopted to simulate the interaction result of the rollouts with the environment, and generate trajectories that minimize contact force. In qualitative and quantitative experiments, SVG-MPPI outperforms the existing paradigm that uses only binary movability for planning, achieving higher success rates with reduced cumulative contact forces. Our code is available at: https://github.com/tud-amr/SVG-MPPI

cs.RO↗

arXiv2025

Damping Tuning Considering Random Disturbances Adopting Distributionally Robust Optimization

In scenarios where high penetration of renewable energy sources (RES) is connected to the grid over long distances, the output of RES exhibits significant fluctuations, making it difficult to accurately characterize. The intermittency and uncertainty of these fluctuations pose challenges to the stability of the power system. This paper proposes a distributionally robust damping optimization control framework (DRDOC) to address the uncertainty in the true distribution of random disturbances caused by RES. First, the installation location of damping controllers and key control parameters are determined through Sobol sensitivity indices and participation factors. Next, a nonlinear relationship between damping and random disturbances is established with Polynomial Chaos Expansion (PCE). The uncertainty in the distribution of disturbances is captured by ambiguity sets. The DRDOC is formulated as a convex optimization problem, which is further simplified for efficient computation. Finally, the optimal control parameters are derived through convex optimization techniques. Simulation results demonstrate the effectiveness and distribution robustness of the proposed DRDOC.

math.OC↗

arXiv2025

Mechanisms and Scale-up Potential of 3D Solar Interfacial-Evaporators

Evaporation rates from porous evaporators under sunlight have been reported to exceed the solar-thermal limit, determined by relating the incoming solar energy to the latent and sensible heat of water, for applications in desalination and brine pond drying. Although flat two-dimensional (2D) evaporators exceeding the solar limit implies a non-thermal process, tall three-dimensional (3D) solar evaporators can exceed it by absorbing additional environmental heat into its cold sidewalls. Through modeling, we explain the physics and identify the critical heights in which a fin transitions from 2D to 3D evaporation and exceeds the solar-thermal limit. Our analyses illustrate that environmental heat absorption in 3D evaporators is determined by the ambient relative humidity and the airflow velocity. The model is then coarse-grained into a large-scale fin array device on the meters scale to analyze their scalability. We identify that these devices are unlikely to scale favorably in closed environment settings such as solar stills. Our modeling clearly illustrates the benefits and limitations of 3D evaporating arrays and pinpoints design choices in previous works that hinder the device's overall performance. This work illustrates the importance in distinguishing 2D from 3D evaporation for mechanisms underlying interfacial evaporation exceeding the solar-thermal limit.

physics.app-ph↗

arXiv2025

Large Language Model for Lossless Image Compression with Visual Prompts

Recent advancements in deep learning have driven significant progress in lossless image compression. With the emergence of Large Language Models (LLMs), preliminary attempts have been made to leverage the extensive prior knowledge embedded in these pretrained models to enhance lossless image compression, particularly by improving the entropy model. However, a significant challenge remains in bridging the gap between the textual prior knowledge within LLMs and lossless image compression. To tackle this challenge and unlock the potential of LLMs, this paper introduces a novel paradigm for lossless image compression that incorporates LLMs with visual prompts. Specifically, we first generate a lossy reconstruction of the input image as visual prompts, from which we extract features to serve as visual embeddings for the LLM. The residual between the original image and the lossy reconstruction is then fed into the LLM along with these visual embeddings, enabling the LLM to function as an entropy model to predict the probability distribution of the residual. Extensive experiments on multiple benchmark datasets demonstrate our method achieves state-of-the-art compression performance, surpassing both traditional and learning-based lossless image codecs. Furthermore, our approach can be easily extended to images from other domains, such as medical and screen content images, achieving impressive performance. These results highlight the potential of LLMs for lossless image compression and may inspire further research in related directions.

eess.IV↗

arXiv2025

Photomolecular Effect as A Potential Explanation for The Cloud Absorption Anomaly

Cloud absorption is acknowledged as the biggest source of uncertainty in the climate models. For over 70 years, many experiments have reported clouds absorbing more solar radiation than theory could predict. In the visible spectrum, simulations based on optical constants of water lead to negligible cloud absorption. This result had been explored by some experimentalists to calibrate the cloud absorption measurements. However, the author and his collaborators recently discovered that visible light can directly cleave off water molecular clusters at liquid-air interfaces (PNAS, e2312751120, 2023; e2320844121, 2024), which is named the photomolecular effect in analogy to the photoelectric effect. This discovery suggests that a crucial piece of physics has been missing in the existing theories: light can be absorbed at water-air interface. The photomolecular effect can be simulated by generalizing the boundary conditions for the Maxwell equations using Feibelman parameters that were derived in the past research on the surface photoelectric effect and surface plasmons. In this work, the author uses simulation to show that including the photomolecular effect at the air-water interface can potentially explain the cloud absorption anomaly. Although the lack of accurate Feibelman parameter values prevents a direct comparison of the current theory with experiments, this work points to an unexplored mechanism for explaining the cloud absorption anomaly and calls for further investigation on the impacts of photomolecular effect in the climate modeling.

physics.ao-ph↗

arXiv2025

Collaborative design of fault diagnosis and fault tolerance control under nested signal temporal logic specifications

Signal Temporal Logic (STL) specifications play a crucial role in defining complex temporal properties and behaviors in safety-critical cyber-physical systems (CPS). However, fault diagnosis (FD) and fault-tolerant control (FTC) for CPS with nonlinear dynamics remain significant challenges, particularly when dealing with nested signal temporal logic (NSTL) specifications. This paper introduces a novel framework for the collaborative design of FD and FTC, aimed at optimizing fault diagnostic performance while ensuring fault tolerance under NSTL specifications. The proposed framework consists of four key steps: (1) construction of the Signal Temporal Logic Tree (STLT), (2) fault detection via the construction of fault-tolerant feasible sets, (3) evaluation of fault detection performance, and (4) synthesis of fault-tolerant control. Initially, a controller for nonlinear systems is designed to satisfy NSTL specifications, and a fault detection observer is developed alongside fault-tolerant feasible sets. To address the challenge of maintaining solution feasibility in dynamic optimization control problems, the concept of fault-tolerant control recursive feasibility is introduced. Subsequently, suboptimal controller gains are derived through a quadratic programming approach to ensure fault tolerance. The collaborative design framework enables more rapid and accurate fault detection while preserving FTC performance. A simulation study is presented to demonstrate the effectiveness of the proposed framework.

eess.SY↗

arXiv2025

Exploit Gradient Skewness to Circumvent Byzantine Defenses for Federated Learning

Federated Learning (FL) is notorious for its vulnerability to Byzantine attacks. Most current Byzantine defenses share a common inductive bias: among all the gradients, the densely distributed ones are more likely to be honest. However, such a bias is a poison to Byzantine robustness due to a newly discovered phenomenon in this paper - gradient skew. We discover that a group of densely distributed honest gradients skew away from the optimal gradient (the average of honest gradients) due to heterogeneous data. This gradient skew phenomenon allows Byzantine gradients to hide within the densely distributed skewed gradients. As a result, Byzantine defenses are confused into believing that Byzantine gradients are honest. Motivated by this observation, we propose a novel skew-aware attack called STRIKE: first, we search for the skewed gradients; then, we construct Byzantine gradients within the skewed gradients. Experiments on three benchmark datasets validate the effectiveness of our attack

cs.LG↗

arXiv2025

3D-Grounded Vision-Language Framework for Robotic Task Planning: Automated Prompt Synthesis and Supervised Reasoning

Vision-language models (VLMs) have achieved remarkable success in scene understanding and perception tasks, enabling robots to plan and execute actions adaptively in dynamic environments. However, most multimodal large language models lack robust 3D scene localization capabilities, limiting their effectiveness in fine-grained robotic operations. Additionally, challenges such as low recognition accuracy, inefficiency, poor transferability, and reliability hinder their use in precision tasks. To address these limitations, we propose a novel framework that integrates a 2D prompt synthesis module by mapping 2D images to point clouds, and incorporates a small language model (SLM) for supervising VLM outputs. The 2D prompt synthesis module enables VLMs, trained on 2D images and text, to autonomously extract precise 3D spatial information without manual intervention, significantly enhancing 3D scene understanding. Meanwhile, the SLM supervises VLM outputs, mitigating hallucinations and ensuring reliable, executable robotic control code generation. Our framework eliminates the need for retraining in new environments, thereby improving cost efficiency and operational robustness. Experimental results that the proposed framework achieved a 96.0\% Task Success Rate (TSR), outperforming other methods. Ablation studies demonstrated the critical role of both the 2D prompt synthesis module and the output supervision module (which, when removed, caused a 67\% TSR drop). These findings validate the framework's effectiveness in improving 3D recognition, task planning, and robotic task execution.

cs.RO↗

arXiv2025

Influences of accretion flow and dilaton charge on the images of Einstein-Maxwell-dilation black hole

The characteristics and images of Einstein-Maxwell-Dilaton (EMD) black holes are examined in this paper, focusing on their effective potential, photon trajectories, and images with thin and thick accretion disks. We found that the shadow and photon sphere radii decrease with increasing dilaton charge. As the observation inclination increases, direct and secondary images become separate, with the direct image appearing hat-shaped. Simulations indicate that the brightness of the shadow and photon ring is higher in static spherical accretion flows compared to infalling ones. The study also shows that in thin disk accretion flows, the direct emission predominantly influences observed luminosity, with photon ring emission being less significant. Additionally, the appearance of black hole images varies with the observer's inclination angle.

gr-qc↗

arXiv2025

Interfacial Temperature and Density Discontinuities for Phase-Change Heat Transfer With Non-condensable Gas

In recent prior work, the author derived interfacial mass and heat flux conditions for phase-change processes. The mass flux condition is identical to the Schrage equation, but the additional heat flux expression enables one to couple the interface to the continua in both the liquid and the vapor phases and compute the interfacial temperature and density discontinuities. However, questions exist on how to treat phase change heat transfer in the presence of non-condensable gases. In this work, the author shows that the same set of interfacial conditions can be used to account for the presence of non-condensable gases. Although the mass flux of non-condensable gas is zero, their presence impacts the heat transfer. For evaporation, when the presence of the non-condensable gas is small, temperature and density discontinuities persist across the interface, as well as inverted temperature distributions. For condensation, however, no temperature inversion happens in the presence of a small amount of non-condensable gas and the interfacial temperature jump is significantly smaller. When a large amount of non-condensable gas is present, such as for evaporation into and condensation from air, the temperature discontinuities at the interface are significantly smaller and no temperature inversion happens. For evaporation driven purely by humidity difference, temperature inversion and discontinuity still exist. Results from this work will benefit the modeling of phase change processes in the presence of non-condensable gases, evaporative cooling in air, air-gap distillation, atmospheric water harvesting, and other applications.

physics.app-ph↗

arXiv2025

Multilineage-differentiating stress-enduring cells alleviate neuropathic pain in mice by secreting TGF-b and IL-10

Neuropathic pain is a chronic condition characterized by damage to and dysfunction of the peripheral or central nervous system. There are currently no effective treatment options available for neuropathic pain, and existing drugs often provide only temporary relief with potential side effects. Multilineage-differentiating stress-enduring (Muse) cells are characterized by high expansion potential, a stable phenotype and strong immunosuppression. These properties make them attractive candidates for therapeutics for neuropathic pain management. In this study, we conducted a series of experiments to evaluate the effect of Muse cells on neuropathic pain. Muse cells from different species demonstrated analgesic potential by reversing CCI-induced neuropathic pain. Protein profiling revealed a high degree of similarity between Muse cells and BMSCs. The intrathecal injection of Muse cells effectively reduced neuropathic pain in various mouse models, resulting in better analgesic effects than the administration of equivalent low doses of BMSCs. Immunohistochemical analysis and qPCR revealed the ability of Muse cells to inhibit spinal cord neuroinflammation caused by SNI. In addition, Transwell and ELISA revealed that Muse cells migrated through the injured dorsal root ganglion (DRG) via the CCR7-CCL21 chemotactic axis. In addition, the secretion of TGF-b and IL-10 by Muse cells was identified as the mechanism underlying the analgesic effect of Muse cells. The capacity of Muse cells to mitigate neuroinflammation and produce analgesic effects via the modulation of TGF-b and IL-10 underscores their potential as promising therapeutic approaches for the treatment of neuropathic pain.

q-bio.TO↗

arXiv2025

Theory of the Photomolecular Effect

It is well-known that water in both liquid and vapor phases exhibits exceptionally weak absorption of light in the visible range. Recent experiments, however, have demonstrated that at the liquid-air interface, absorption in the visible range is drastically increased. This increased absorption results in a rate of evaporation that exceeds the theoretical thermal limit by between two and five times. Curiously, the evaporation rate peaks at green wavelengths of light, while no corresponding absorptance peak has been observed. Experiments suggest that photons can cleave off clusters of water molecules at the surface, but no clear theoretical model has yet been proposed to explain how this is possible. This paper aims to present such a model and explain this surprising and important phenomenon.

cond-mat.soft↗

arXiv2025

NeurDB: On the Design and Implementation of an AI-powered Autonomous Database

Databases are increasingly embracing AI to provide autonomous system optimization and intelligent in-database analytics, aiming to relieve end-user burdens across various industry sectors. Nonetheless, most existing approaches fail to account for the dynamic nature of databases, which renders them ineffective for real-world applications characterized by evolving data and workloads. This paper introduces NeurDB, an AI-powered autonomous database that deepens the fusion of AI and databases with adaptability to data and workload drift. NeurDB establishes a new in-database AI ecosystem that seamlessly integrates AI workflows within the database. This integration enables efficient and effective in-database AI analytics and fast-adaptive learned system components. Empirical evaluations demonstrate that NeurDB substantially outperforms existing solutions in managing AI analytics tasks, with the proposed learned components more effectively handling environmental dynamism than state-of-the-art approaches.

cs.DB↗

arXiv2025

Particle-based Instance-aware Semantic Occupancy Mapping in Dynamic Environments

Representing the 3D environment with instance-aware semantic and geometric information is crucial for interaction-aware robots in dynamic environments. Nevertheless, creating such a representation poses challenges due to sensor noise, instance segmentation and tracking errors, and the objects' dynamic motion. This paper introduces a novel particle-based instance-aware semantic occupancy map to tackle these challenges. Particles with an augmented instance state are used to estimate the Probability Hypothesis Density (PHD) of the objects and implicitly model the environment. Utilizing a State-augmented Sequential Monte Carlo PHD (S$^2$MC-PHD) filter, these particles are updated to jointly estimate occupancy status, semantic, and instance IDs, mitigating noise. Additionally, a memory module is adopted to enhance the map's responsiveness to previously observed objects. Experimental results on the Virtual KITTI 2 dataset demonstrate that the proposed approach surpasses state-of-the-art methods across multiple metrics under different noise conditions. Subsequent tests using real-world data further validate the effectiveness of the proposed approach.

cs.RO↗

arXiv2025

Mining Platoon Patterns from Traffic Videos

Discovering co-movement patterns from urban-scale video data sources has emerged as an attractive topic. This task aims to identify groups of objects that travel together along a common route, which offers effective support for government agencies in enhancing smart city management. However, the previous work has made a strong assumption on the accuracy of recovered trajectories from videos and their co-movement pattern definition requires the group of objects to appear across consecutive cameras along the common route. In practice, this often leads to missing patterns if a vehicle is not correctly identified from a certain camera due to object occlusion or vehicle mis-matching. To address this challenge, we propose a relaxed definition of co-movement patterns from video data, which removes the consecutiveness requirement in the common route and accommodates a certain number of missing captured cameras for objects within the group. Moreover, a novel enumeration framework called MaxGrowth is developed to efficiently retrieve the relaxed patterns. Unlike previous filter-and-refine frameworks comprising both candidate enumeration and subsequent candidate verification procedures, MaxGrowth incurs no verification cost for the candidate patterns. It treats the co-movement pattern as an equivalent sequence of clusters, enumerating candidates with increasing sequence length while avoiding the generation of any false positives. Additionally, we also propose two effective pruning rules to efficiently filter the non-maximal patterns. Extensive experiments are conducted to validate the efficiency of MaxGrowth and the quality of its generated co-movement patterns. Our MaxGrowth runs up to two orders of magnitude faster than the baseline algorithm. It also demonstrates high accuracy in real video dataset when the trajectory recovery algorithm is not perfect.

cs.CV↗

arXiv2024

Cost-Aware Dynamic Cloud Workflow Scheduling using Self-Attention and Evolutionary Reinforcement Learning

The Cost-aware Dynamic Multi-Workflow Scheduling (CDMWS) in the cloud is a kind of cloud workflow management problem, which aims to assign virtual machine (VM) instances to execute tasks in workflows so as to minimize the total costs, including both the penalties for violating Service Level Agreement (SLA) and the VM rental fees. Powered by deep neural networks, Reinforcement Learning (RL) methods can construct effective scheduling policies for solving CDMWS problems. Traditional policy networks in RL often use basic feedforward architectures to separately determine the suitability of assigning any VM instances, without considering all VMs simultaneously to learn their global information. This paper proposes a novel self-attention policy network for cloud workflow scheduling (SPN-CWS) that captures global information from all VMs. We also develop an Evolution Strategy-based RL (ERL) system to train SPN-CWS reliably and effectively. The trained SPN-CWS can effectively process all candidate VM instances simultaneously to identify the most suitable VM instance to execute every workflow task. Comprehensive experiments show that our method can noticeably outperform several state-of-the-art algorithms on multiple benchmark CDMWS problems.

cs.AI↗

arXiv2024

Dynamics of Spinning Binary at 2PM

We consider the covariant proposal for the gravitational Compton amplitude for a Kerr black hole. Employing the covariant three- and four-point Compton amplitudes, we assemble the classical one-loop integrand on the maximal cut at all orders in spin, utilizing the method of unitarity. Expanding in powers of spin, we evaluate the one-loop amplitude up to $\mathcal O(G^2 a^8)$. Supplemented with extra contact contributions derived from the far-zone data of the Teukolsky solutions, the one-loop amplitude is in agreement with results available in the literature. We showcase the classical eikonal in the aligned-spin case at $\mathcal O(G^2 a^7)$.

hep-th↗

arXiv2024

Learning Monocular Depth from Events via Egomotion Compensation

Event cameras are neuromorphically inspired sensors that sparsely and asynchronously report brightness changes. Their unique characteristics of high temporal resolution, high dynamic range, and low power consumption make them well-suited for addressing challenges in monocular depth estimation (e.g., high-speed or low-lighting conditions). However, current existing methods primarily treat event streams as black-box learning systems without incorporating prior physical principles, thus becoming over-parameterized and failing to fully exploit the rich temporal information inherent in event camera data. To address this limitation, we incorporate physical motion principles to propose an interpretable monocular depth estimation framework, where the likelihood of various depth hypotheses is explicitly determined by the effect of motion compensation. To achieve this, we propose a Focus Cost Discrimination (FCD) module that measures the clarity of edges as an essential indicator of focus level and integrates spatial surroundings to facilitate cost estimation. Furthermore, we analyze the noise patterns within our framework and improve it with the newly introduced Inter-Hypotheses Cost Aggregation (IHCA) module, where the cost volume is refined through cost trend prediction and multi-scale cost consistency constraints. Extensive experiments on real-world and synthetic datasets demonstrate that our proposed framework outperforms cutting-edge methods by up to 10\% in terms of the absolute relative error metric, revealing superior performance in predicting accuracy.

cs.CV↗