arXiv subjects

Long Chen

Publications and source records attributed to Long Chen.

At least 253 records · Page 14Linked to original sources

arXiv2025

Team Xiaomi EV-AD VLA: Caption-Guided Retrieval System for Cross-Modal Drone Navigation -- Technical Report for IROS 2025 RoboSense Challenge Track 4

Cross-modal drone navigation remains a challenging task in robotics, requiring efficient retrieval of relevant images from large-scale databases based on natural language descriptions. The RoboSense 2025 Track 4 challenge addresses this challenge, focusing on robust, natural language-guided cross-view image retrieval across multiple platforms (drones, satellites, and ground cameras). Current baseline methods, while effective for initial retrieval, often struggle to achieve fine-grained semantic matching between text queries and visual content, especially in complex aerial scenes. To address this challenge, we propose a two-stage retrieval refinement method: Caption-Guided Retrieval System (CGRS) that enhances the baseline coarse ranking through intelligent reranking. Our method first leverages a baseline model to obtain an initial coarse ranking of the top 20 most relevant images for each query. We then use Vision-Language-Model (VLM) to generate detailed captions for these candidate images, capturing rich semantic descriptions of their visual content. These generated captions are then used in a multimodal similarity computation framework to perform fine-grained reranking of the original text query, effectively building a semantic bridge between the visual content and natural language descriptions. Our approach significantly improves upon the baseline, achieving a consistent 5\% improvement across all key metrics (Recall@1, Recall@5, and Recall@10). Our approach win TOP-2 in the challenge, demonstrating the practical value of our semantic refinement strategy in real-world robotic navigation scenarios.

cs.RO↗

arXiv2025

All-sky search for individual Primordial Black Hole bursts with LHAASO

Primordial Black Holes~(PBHs) are hypothetical black holes with a wide range of masses that formed in the early universe. As a result, they may play an important cosmological role and provide a unique probe of the early universe. A PBH with an initial mass of approximately $10^{15}$~g is expected to explode today in a final burst of Hawking radiation. In this work, we conduct an all-sky search for individual PBH burst events using the data collected from March 2021 to July 2024 by the Water Cherenkov Detector Array of the Large High Altitude Air Shower Observatory (LHAASO). Three PBH burst durations, 10~s, 20~s, and 100~s, are searched, with no significant PBH bursts observed. The upper limit on the local PBH burst rate density is set to be as low as 181~pc$^{-3}$~yr$^{-1}$ at 99$\%$ confidence level, representing the most stringent limit achieved to date.

astro-ph.HE↗

arXiv2025

Decoupling Contrastive Decoding: Robust Hallucination Mitigation in Multimodal Large Language Models

Although multimodal large language models (MLLMs) exhibit remarkable reasoning capabilities on complex multimodal understanding tasks, they still suffer from the notorious hallucination issue: generating outputs misaligned with obvious visual or factual evidence. Currently, training-based solutions, like direct preference optimization (DPO), leverage paired preference data to suppress hallucinations. However, they risk sacrificing general reasoning capabilities due to the likelihood displacement. Meanwhile, training-free solutions, like contrastive decoding, achieve this goal by subtracting the estimated hallucination pattern from a distorted input. Yet, these handcrafted perturbations (e.g., add noise to images) may poorly capture authentic hallucination patterns. To avoid these weaknesses of existing methods, and realize robust hallucination mitigation (i.e., maintaining general reasoning performance), we propose a novel framework: Decoupling Contrastive Decoding (DCD). Specifically, DCD decouples the learning of positive and negative samples in preference datasets, and trains separate positive and negative image projections within the MLLM. The negative projection implicitly models real hallucination patterns, which enables vision-aware negative images in the contrastive decoding inference stage. Our DCD alleviates likelihood displacement by avoiding pairwise optimization and generalizes robustly without handcrafted degradation. Extensive ablations across hallucination benchmarks and general reasoning tasks demonstrate the effectiveness of DCD, i.e., it matches DPO's hallucination suppression while preserving general capabilities and outperforms the handcrafted contrastive decoding methods.

cs.LG↗

arXiv2025

NanoVLA: Routing Decoupled Vision-Language Understanding for Nano-sized Generalist Robotic Policies

Vision-language-action (VLA) models have significantly advanced robotic manipulation by integrating vision-language models (VLMs), and action decoders into a unified architecture. However, their deployment on resource-constrained edge devices, such as mobile robots or embedded systems (e.g., Jetson Orin Nano), remains challenging due to high computational demands, especially in real-world scenarios where power, latency, and computational resources are critical. To close this gap, we introduce Nano-scale Vision-Language Action (NanoVLA), a family of lightweight VLA architectures that achieve high performance with minimal resources. Our core innovations include: (1) vision-language decoupling that moves conventional early vision and language inputs fusion in VLM to late stage, achieving better performance while enabling caching and reduce inference overhead and latency; (2) long-short action chunking to ensure smooth, coherent multi-step planning without sacrificing real-time responsiveness; (3) dynamic routing that adaptively assigns lightweight or heavy backbones based on task complexity, further optimizing inference efficiency. Experimental results on several benchmarks, as well as real-world deployments, demonstrate that NanoVLA achieves up to 52x faster inference on edge devices compared to previous state-of-the-art VLA models, with 98% less parameters while maintaining or surpassing their task accuracy and generalization. Ablation studies confirm that our decoupling strategy preserves cross-task transferability, and the routing module enhances cost-performance trade-offs, enabling practical, high-precision robotic manipulation on resource-constrained hardware.

cs.RO↗

arXiv2025

CoMo: Compositional Motion Customization for Text-to-Video Generation

While recent text-to-video models excel at generating diverse scenes, they struggle with precise motion control, particularly for complex, multi-subject motions. Although methods for single-motion customization have been developed to address this gap, they fail in compositional scenarios due to two primary challenges: motion-appearance entanglement and ineffective multi-motion blending. This paper introduces CoMo, a novel framework for $\textbf{compositional motion customization}$ in text-to-video generation, enabling the synthesis of multiple, distinct motions within a single video. CoMo addresses these issues through a two-phase approach. First, in the single-motion learning phase, a static-dynamic decoupled tuning paradigm disentangles motion from appearance to learn a motion-specific module. Second, in the multi-motion composition phase, a plug-and-play divide-and-merge strategy composes these learned motions without additional training by spatially isolating their influence during the denoising process. To facilitate research in this new domain, we also introduce a new benchmark and a novel evaluation metric designed to assess multi-motion fidelity and blending. Extensive experiments demonstrate that CoMo achieves state-of-the-art performance, significantly advancing the capabilities of controllable video generation. Our project page is at https://como6.github.io/.

cs.CV↗

arXiv2025

A Giant Peanut-shaped Ultra-High-Energy Gamma-Ray Emitter Off the Galactic Plane

Ultra-high-energy (UHE), exceeding 100 TeV (10^12 electronvolts), γ-rays manifests extreme particle acceleration in astrophysical sources. Recent observations by γ-ray telescopes, particularly by the Large High Altitude Air Shower Observatory (LHAASO), have revealed a few tens of UHE sources, indicating numerous Galactic sources capable of accelerating particles to PeV (10^15 electronvolts) energies. However, discerning the dominant acceleration mechanisms (leptonic versus hadronic), the relative contributions of specific source classes, and the role of particle transport in shaping their observed emission are central goals of modern UHE astrophysics. Here we report the discovery of a giant UHE γ-ray emitter at -17.5° off the Galactic plane - a region where UHE γ-ray sources are rarely found. The emitter exhibits a distinctive asymmetric shape, resembling a giant "Peanut" spanning 0.45° \times 4.6°, indicative of anisotropic particle distribution over a large area. A highly aged millisecond pulsar (MSP) J0218+4232 is the sole candidate accelerator positionally coincident with the Peanut region. Its association with UHE γ-rays extending to 0.7 PeV, if confirmed, would provide the first evidence of a millisecond pulsar powering PeV particles. Such a finding challenges prevailing models, which posit that millisecond pulsars cannot sustain acceleration to PeV energies. The detection reveals fundamental gaps in understanding particle acceleration, cosmic-ray transport, and interstellar magnetic field effects, potentially revealing new PeV accelerator (PeVatron) classes.

astro-ph.HE↗

arXiv2025

Hybridizable Symmetric Stress Elements on the Barycentric Refinement in Arbitrary Dimensions

Hybridizable $H(\textrm{div})$-conforming finite elements for symmetric tensors on simplices with barycentric refinement are developed in this work for arbitrary dimensions and any polynomial order. By employing barycentric refinement and an intrinsic tangential-normal ($t$-$n$) decomposition, novel basis functions are constructed to redistribute degrees of freedom while preserving $H(\textrm{div})$-conformity and symmetry, and ensuring inf-sup stability. These hybridizable elements enhance computational flexibility and efficiency, with applications to mixed finite element methods for linear elasticity.

math.NA↗

arXiv2025

On-shell Matrix Elements of the EMT Trace in Gauge Theories and Heavy Quark Masses

We present a novel diagrammatic proof of the identity between the forward matrix element of the energy-momentum-tensor (EMT) trace operator over a single particle's on-shell state and its perturbative pole mass to any loop orders in perturbative gauge theories (with gauge bosons and fermions), without appealing to any pre-laid operator renormalization conditions or Ward identities. The proof is based on the equation of mass-dimensional analysis in dimensional regularization, the topological properties of contributing Feynman diagrams and the on-shell renormalization condition. Considering for definiteness perturbative QED and QCD with at most one fermion kept massive, we have verified the aforementioned identity, up to three loops, for all elementary particles through direct computation of dimensionally-regularized matrix elements of the relevant bare operators. We observe interestingly that the trace-anomaly contribution seemingly contains all leading-renormalon effects (explicitly verified up to three loops), and we propose accordingly a new scheme- and scale-independent trace-anomaly-subtracted mass definition for heavy $t$-,$\,b$-,$\,c$-quarks and electrons. A list of amusing numbers is subsequently presented for the composition of their perturbative pole masses.

hep-ph↗

arXiv2025

Segmentation as A Plug-and-Play Capability for Frozen Multimodal LLMs

Integrating diverse visual capabilities into a unified model is a significant trend in Multimodal Large Language Models (MLLMs). Among these, the inclusion of segmentation poses a distinct set of challenges. To equip MLLMs with pixel-level segmentation abilities, prevailing methods require finetuning the model to produce specific outputs compatible with a mask decoder. This process typically alters the model's output space and compromises its intrinsic generalization, which undermines the goal of building a unified model. We introduce LENS (Leveraging kEypoiNts for MLLMs' Segmentation), a novel plug-and-play solution. LENS attaches a lightweight, trainable head to a completely frozen MLLM. By refining the spatial cues embedded in attention maps, LENS extracts keypoints and describes them into point-wise features directly compatible with the mask decoder. Extensive experiments validate our approach: LENS achieves segmentation performance competitive with or superior to that of retraining-based methods. Crucially, it does so while fully preserving the MLLM's generalization capabilities, which are significantly degraded by finetuning approaches. As such, the attachable design of LENS establishes an efficient and powerful paradigm for extending MLLMs, paving the way for truly multi-talented, unified models.

cs.CV↗

arXiv2025

A Novel Preconditioning Framework for Solving Nonlinear PDEs based on Fenchel-Rockafellar Duality and Transformed Primal-Dual Techniques

A DualTPD method is proposed for solving nonlinear partial differential equations. The method is characterized by three main features. First, decoupling via Fenchel--Rockafellar duality is achieved, so that nonlinear terms are discretized by discontinuous finite element spaces, yielding block-diagonal mass matrices and closed-form updates. Second, improved convergence is obtained by applying transformed primal--dual (TPD) dynamics to the nonlinear saddle-point system, which yields strongly monotone behavior. Third, efficient preconditioners are designed for the elliptic-type Schur complement arising from the separated differential operators, and multigrid solvers are applied effectively. Extensive numerical experiments on elliptic $p$-Laplacian and nonlinear $H(\curl)$ problems are presented, showing significant efficiency gains with global, mesh-independent convergence.

math.NA↗

arXiv2025

Generative model for information metamaterial design

Generative models such as AlphaFold and MatterGen can directly generate novel material structures with desired properties, accelerating the new materials discovery and revolutionizing the material design paradigm from traditional trial-and-error approach to intelligent on-demand generation. AlphaFold is focused on protein prediction with specific aperiodic structures; while MatterGen is focused on predicting periodic and stable crystal structures. The universal design of metamaterials is much more complicated, since it involves to design meta-atoms (similar to the periodic structures) and their arbitrarily inhomogeneous distributions in space. Here, we propose InfoMetaGen, a universal generative model for information metamaterial design, which combines a pre-trained foundation model with lightweight functional adapters to intelligently generate artificial structures on-demand spanning from meta-atoms to arbitrary space coding patterns. In contrast to conventional intelligent metamaterial design methods that require training dedicated models for specific functionalities, InfoMetaGen enables a single universal generative model capable of switching across diverse functionalities by fine-tuning the lightweight adapters, significantly improving both efficiency and generalizability. Experimental results demonstrate that InfoMetaGen can not only accelerate the diverse discovery of new metamaterials, but also achieve breakthroughs in metamaterial performance. This work fills the gap of universal generative framework in designing artificial materials, and opens up unprecedented opportunities to expand the capability of generative models from the passive discovery of microscopic natural material to the active creation of macroscopic artificial materials.

physics.optics↗

arXiv2025

Stock Prediction via a Dual Relation Fusion Network incorporating Static and Dynamic Relations

Accurate modeling of inter-stock relationships is critical for stock price forecasting. However, existing methods predominantly focus on single-state relationships, neglecting the essential complementarity between dynamic and static inter-stock relations. To solve this problem, we propose a Dual Relation Fusion Network (DRFN) to capture the long-term relative stability of stock relation structures while retaining the flexibility to respond to sudden market shifts. Our approach features a novel relative static relation component that models time-varying long-term patterns and incorporates overnight informational influences. We capture dynamic inter-stock relationships through distance-aware mechanisms, while evolving long-term structures via recurrent fusion of dynamic relations from the prior day with the pre-defined static relations. Experiments demonstrate that our method significantly outperforms the baselines across different markets, with high sensitivity to the co-movement of relational strength and stock price.

cs.LG↗

arXiv2025

Reasoning through Exploration: A Reinforcement Learning Framework for Robust Function Calling

The effective training of Large Language Models (LLMs) for function calling faces a critical challenge: balancing exploration of complex reasoning paths with stable policy optimization. Standard methods like Supervised Fine-Tuning (SFT) fail to instill robust reasoning, and traditional Reinforcement Learning (RL) struggles with inefficient exploration. We propose \textbf{EGPO}, a new RL framework built upon Group Relative Policy Optimization (GRPO), designed to address this challenge directly. The core of EGPO is an entropy-enhanced advantage function that integrates the entropy of the model's Chain-of-Thought (CoT) into the policy gradient computation. This encourages the generation of diverse reasoning strategies. To maintain optimization direction, the entropy bonus is carefully constrained by a clipping mechanism. Complemented by a strict, binary reward signal, EGPO effectively guides the model towards discovering structured and accurate tool invocation patterns. On the challenging Berkeley Function Calling Leaderboard (BFCL), a 4B-parameter model trained with EGPO sets a new state-of-the-art among models of comparable size, surpassing a range of strong competitors, including GPT-4o and Gemini-2.5.

cs.LG↗

arXiv2025

New Physics Search at the CEPC: a General Perspective

The Circular Electron-Positron Collider (CEPC), a proposed next-generation Higgs factory, provides new opportunities to explore physics beyond the Standard Model (SM). With its clean electron-positron collision environment and the ability to collect large samples of Higgs, W, and Z bosons, the CEPC enables precision measurements and searches for new physics. This white paper outlines the CEPC's discovery potential, including studies of exotic decays of the Higgs, Z, and top quarks, dark matter and dark sector phenomena, long-lived particles, supersymmetry, and neutrino-related signatures. Advanced detector technologies and reconstruction techniques, such as one-to-one correspondence reconstruction and jet origin identification, significantly improve sensitivity to rare and weakly interacting processes. The CEPC is particularly well suited to probe the electroweak phase transition and test models of electroweak baryogenesis and dark sector interactions. In addition, global fit analyses highlight the CEPC's complementary role in constraining a wide range of new physics scenarios. These features position the CEPC as a powerful tool for exploring the next frontier in fundamental particle physics in the post-Higgs discovery era.

hep-ex↗

arXiv2025

Determination of the Strong Coupling Constant $α_s$ from Inclusive Semi-leptonic $B$ Meson Decays

We demonstrate the feasibility of determining the strong coupling constant, $α_s$, from the inclusive semileptonic decay width of $B$ mesons. We express the semileptonic $B$ decay width as a function of $α_s(5\mathrm{\,GeV})$, the Cabibbo-Kobayashi-Maskawa matrix element $|V_{cb}|$, $b$- and $c$-quark masses in the $\overline{\mathrm{MS}}$ scheme. We fit $α_s(5\mathrm{\,GeV})$ to current world averages of the $B^{\pm}$ and $B^{0}$ semileptonic decay widths. This yields $α_s(5\mathrm{\,GeV}) = 0.245 \pm 0.009$, corresponding to a 5-flavor extrapolation of $α_s(m_{Z}) = 0.1266 \pm 0.0023$. The primary uncertainty contributions arise from the uncertainty on the perturbative expansion and the value of $|V_{cb}|$. Future advancements including higher-order perturbative calculations, and precise measurements of $|V_{cb}|$ and $B$ decay widths from upcoming $B$ and $Z$ factories, could enable this method to determine $α_s(m_{Z})$ with a competitive precision of $Δα_s(m_{Z}) \sim 0.0018$. This precision is comparable to the current accuracy of $α_s(m_{Z})$ measurements from $τ$-lepton decays, which is regarded as the most precise experimental approach.

hep-ph↗

arXiv2025

Elastic Energy Storage Mechanism in Hovering Animal Flight: A Discriminative Method Based on Wing Kinematics

Existing research has yet to reach a consensus on whether and how small flying animals utilize elastic energy storage mechanisms to reduce flight energy expenditure, and there is a lack of systematic and universal methods for assessment. To address these gaps, this study proposes a method to evaluate elastic energy storage capacity based on wing kinematic parameters (flapping amplitude and flapping frequency), grounded in the hypothesis that animals tend to minimize flight energy expenditure. By establishing a simplified power model, the study calculates the optimal kinematic parameters corresponding to the minimum mechanical power requirements under two extreme conditions: no elastic energy storage and complete elastic energy storage. These optimal parameters are then compared with measured data from various small flying animals. The results show that the measured parameters of hummingbirds, ladybugs, and rhinoceros beetles are close to the no-storage optimum, indicating relatively weak elastic energy storage capacity; whereas hoverflies, bumblebees, and honeybees align closely with the complete-storage optimum, suggesting strong elastic energy storage ability. Furthermore, the wing kinematic adjustment strategies these animals employ in response to changes in load or air density are consistent with the predicted elastic storage capacities. This study provides a systematic new approach for assessing biological elastic energy storage capacity and offers a theoretical basis for the low-power design of flapping wing micro air vehicles.

physics.bio-ph↗

arXiv2025

ReCogDrive: A Reinforced Cognitive Framework for End-to-End Autonomous Driving

Recent studies have explored leveraging the world knowledge and cognitive capabilities of Vision-Language Models (VLMs) to address the long-tail problem in end-to-end autonomous driving. However, existing methods typically formulate trajectory planning as a language modeling task, where physical actions are output in the language space, potentially leading to issues such as format-violating outputs, infeasible actions, and slow inference speeds. In this paper, we propose ReCogDrive, a novel Reinforced Cognitive framework for end-to-end autonomous Driving, unifying driving understanding and planning by integrating an autoregressive model with a diffusion planner. First, to instill human driving cognition into the VLM, we introduce a hierarchical data pipeline that mimics the sequential cognitive process of human drivers through three stages: generation, refinement, and quality control. Building on this cognitive foundation, we then address the language-action mismatch by injecting the VLM's learned driving priors into a diffusion planner to efficiently generate continuous and stable trajectories. Furthermore, to enhance driving safety and reduce collisions, we introduce a Diffusion Group Relative Policy Optimization (DiffGRPO) stage, reinforcing the planner for enhanced safety and comfort. Extensive experiments on the NAVSIM and Bench2Drive benchmarks demonstrate that ReCogDrive achieves state-of-the-art performance. Additionally, qualitative results across diverse driving scenarios and DriveBench highlight the model's scene comprehension. All code, model weights, and datasets will be made publicly available to facilitate subsequent research.

cs.CV↗

arXiv2025

DiTraj: training-free trajectory control for video diffusion transformer

Diffusion Transformers (DiT)-based video generation models with 3D full attention exhibit strong generative capabilities. Trajectory control represents a user-friendly task in the field of controllable video generation. However, existing methods either require substantial training resources or are specifically designed for U-Net, do not take advantage of the superior performance of DiT. To address these issues, we propose DiTraj, a simple but effective training-free framework for trajectory control in text-to-video generation, tailored for DiT. Specifically, first, to inject the object's trajectory, we propose foreground-background separation guidance: we use the Large Language Model (LLM) to convert user-provided prompts into foreground and background prompts, which respectively guide the generation of foreground and background regions in the video. Then, we analyze 3D full attention and explore the tight correlation between inter-token attention scores and position embedding. Based on this, we propose inter-frame Spatial-Temporal Decoupled 3D-RoPE (STD-RoPE). By modifying only foreground tokens' position embedding, STD-RoPE eliminates their cross-frame spatial discrepancies, strengthening cross-frame attention among them and thus enhancing trajectory control. Additionally, we achieve 3D-aware trajectory control by regulating the density of position embedding. Extensive experiments demonstrate that our method outperforms previous methods in both video quality and trajectory controllability.

cs.CV↗