Searcharxiv⌕ Search

arXiv subjects

Xin Zhou

Publications and source records attributed to Xin Zhou.

At least 19 recordsLinked to original sources
arXiv2026

Qwen-Drive-1.0: An Initial Step towards a Vision-Language Foundation Model for Autonomous Driving

We present Qwen-Drive-1.0, an initial step towards a vision-language foundation model for autonomous driving. Qwen-Drive-1.0 retains the architecture of the pretrained vision-language model (VLM) and integrates 3D perception, visual question answering, and motion planning within a unified framework. An external bird's-eye-view (BEV) perception head jointly performs 3D object detection, semantic occupancy prediction, and BEV map segmentation. It serves as a probe of the 3D information accessible from the shared representations and provides an explicit, inspectable interface to 3D scene structure. A Planning Expert conditions on shared VLM representations to generate future ego trajectories. A staged training recipe combines driving supervision with general-purpose vision-language data to acquire driving-specific competence while helping preserve broad visual understanding and instruction-following capabilities. Experiments demonstrate strong 3D perception and driving scene understanding while largely preserving general vision-language capability. Comprehensive evaluations across open-loop, pseudo-closed-loop, and closed-loop settings further show highly competitive motion-planning performance.

cs.CV↗
arXiv2026

CGFM-Nav: Cognitive Graph-Field Memory for Semantic-Guided Lifelong Multimodal Embodied Navigation

Vision-and-Language Navigation (VLN) requires agents to reason over accumulated observations while continuously exploring unseen regions. However, existing environment representations often struggle to jointly support explicit semantic memory and continuous exploration guidance. To address this challenge, we propose Cognitive Graph-Field Memory (CGFM), a persistent multimodal scene representation that couples explicit relational memory with continuous spatial intuition. CGFM organizes objects, spatial relations, and visual observations into a multimodal scene graph, enabling target retrieval and long-horizon reasoning across navigation tasks. When no reliable target match is identified, graph-based evidence is projected into a goal-conditioned semantic-frontier field to guide exploration toward semantically promising frontiers and regions. Building upon CGFM, we introduce CGFM-Nav, a foundation-model-based framework for lifelong multimodal navigation that integrates task-relevant subgraph selection, VLM reasoning, and verification feedback into a closed decision loop. Preliminary experiments on GOAT-Bench show that, under the same Qwen3-VL-8B backbone, CGFM-Nav improves the overall success rate from 53.2% to 63.0% and SPL from 30.0% to 39.6%, demonstrating the effectiveness of combining explicit semantic memory with semantic-guided exploration.

cs.RO↗
arXiv2026

Dark-Mode Control of Contrasting Entanglement and Bell Nonlocality between Mechanical Oscillators

This study presents a detailed proposal for an optomechanical system consisting of two mechanical oscillators coupled to a common cavity, aimed at generating pure and entangled two-mode squeezed mechanical steady states. We found that the violation of Bell's measurement may not occur where the entanglement is maximum; rather, nonlocality can be observed for lower entangled states. A central result is that optomechanical coupling imperfections can enhance mechanical entanglement while simultaneously suppressing Bell nonlocality by reducing the purity of the mechanical state. To mitigate this trade-off, we introduce phase-dependent phonon hopping between the mechanical oscillators and show that Bell nonlocality can be selectively enhanced in specific dark-mode configurations, even when the overall entanglement is reduced. We trace this contrasting behaviorto changes in state purity associated with the imbalance of the Bogoliubov-mode occupations. Compatible with existing microwave cavity optomechanical platforms, the proposed architecture provides an experimentally accessible route for controlling nonlocal quantum correlations in multimode mechanical systems. Our proposed scheme serves as an attractive platform for the deployment of continuous-variable teleportation and high-fidelity quantum communication.

quant-ph↗
arXiv2026

Cusp-singularity-enhanced Coriolis effect for ultrasensitive chip-scale gyroscopes

Gyroscopes, as fundamental inertial sensors, are crucial for rotation measurements in consumer electronics, automotive, and aerospace industries, with the most widely used kind relying on the Coriolis effect. The chip-scale Coriolis vibratory gyroscopes (CVGs) show reduced size, weight, and cost, but remain far lower performance than traditional macroscale CVGs, as the weak intrinsic Coriolis factor sets a fundamental limit on scaling the sensitivity against the inherently louder Brownian noise in microchips compared to the macroscale ones. Here, to overcome this physical limit, for the first time, we propose and experimentally demonstrate the use of third-order singularities lying within cusp catastrophes in the phase-tracked oscillations of an on-chip CVG to facilitate a cubic-root scaling of the Coriolis-effect-induced frequency modulation. Employing this effect, we achieve a three-order-of-magnitude enhancement in the Coriolis factor, yielding a 253-fold improvement in signal-to-noise ratio and a 297-fold increase in precision. Moreover, the cusp singularity enables a previously unattainable ultrasensitive phase-modulated sublinear measurement, achieving a world-record signal-to-noise ratio performance for silicon-chip gyroscopes. These findings not only provide revolutionary advancements in gyroscope technologies, by filling the gap in observing and controlling the singularity-enhanced Coriolis effect, but also shed new light on other ultrasensitive sensing applications.

physics.ins-det↗
arXiv2026

A Unified Issue Resolution Benchmark for Requirement Clarification, Planning, and Code Generation for Coding Agents

Large language model-powered coding agents are increasingly used to modify existing code repositories, for example, by adding features or fixing bugs. Yet existing repository-level benchmarks typically evaluate only whether the final patch passes tests. Satisfying a user request requires a long chain of interdependent reasoning and decisions: an agent must recover explicit and implicit requirements, formulate a repository-grounded implementation plan, and translate it into correct code. A pass/fail outcome cannot characterize how an unsuccessful trajectory diverges from the requirements and implementation process needed for a correct patch. To address this gap, we introduce SWE-RPG, a repository-level benchmark that combines executable patch evaluation with validated ground-truth references (GTs) for (1) Requirement Clarification and (2) Implementation Planning. These intermediate GTs support retrospective, GT-aligned diagnosis of complete coding-agent trajectories across clarification, planning, code generation, and artifact submission. SWE-RPG comprises 163 tasks from 31 Python and Java repositories, including 113 bug fixes and 50 feature additions. We evaluate 3 coding agents, including Claude Code, Codex, and OpenCode, with 6 large language model backends, including Claude-Sonnet-5 and GPT-5.6-Terra. Results show that the evaluated popular coding agents still struggle to implement user requests in existing repositories, achieving an average resolved rate of only 31.5% on SWE-RPG. Intermediate-GT diagnosis further identifies implicit requirement recovery as the main bottleneck, accounting for 24.5%--46.0% of agent runs. This result suggests implicit-requirement recovery as a key candidate direction for improving coding agents. The benchmark data and evaluation code are available at https://github.com/Xin-Zhou-smu/SWE-RPG-Bench.

cs.SE↗
arXiv2026

SimWAM: A Simple World Action Model for End-to-End Autonomous Driving

World-Action Models (WAMs) improve end-to-end autonomous driving by transferring video dynamics priors to action prediction, but existing methods incur costly test-time future imagination. We present SimWAM, a simple yet effective WAM that leverages future-video prediction as a training-time supervision signal. It co-trains a pretrained video expert and a lightweight action expert with joint flow matching. An isolated attention mask keeps action prediction independent of future frames, allowing trajectory prediction without explicit future-frame generation at inference. Since the two experts share no parameters and interact only through a unified attention interface, the video backbone could be replaced and the action expert scaled independently without modifying the learning objective or inference pipeline. We further apply reinforcement learning to optimize a compositional driving reward beyond trajectory imitation. Our SimWAM achieves 91.5 PDMS on NAVSIM, surpasses state-of-the-art WAM-based planners with substantially lower latency, and transfers zero-shot to nuScenes. These results position SimWAM as a simple yet solid baseline that could readily benefit from advances in video generation for efficient autonomous driving. The code and model weights are available at https://github.com/H-EmbodVis/SimWAM/.

cs.CV↗
arXiv2026

CO Structures with Narrow Lines in Nearby Quiescent Regions

Using CO data from Phase I of the Milky Way Imaging Scroll Painting (MWISP) survey, we present a systematic study of molecular structures with narrow lines. We identify 57 CO structures, most of which exhibit low densities and subsonic/transonic turbulence. Among them, structures with large projected areas and diffuse, sheet-like geometries are identified as veil clouds. The low LSR velocities and the concentration of these CO structures toward both the Galactic center (e.g., Ophiuchus, Aquila) and anticenter (e.g., Cepheus, Taurus) regions suggest a local origin for the sample, as supported by distance measurements of about 200--300pc for a subset with relatively large angular extents. These nearby structures likely arise from large-scale compression driven by past supernova activity within the Local Bubble. The observed low-velocity-dispersion emission may trace quiescent regions where turbulence has decayed due to a lack of sustained energy injection. For diffuse veil clouds with an assumed magnetic field of ~10uG, ion-neutral friction may provide an additional mechanism for turbulent dissipation on sub-parsec scales corresponding to their thickness of 0.1--0.3pc. Tracing the atomic-to-molecular transition, veil clouds provide a unique window into the diffuse, quiescent precursor state of dense gas. They likely represent a widespread but previously overlooked component of the Galactic molecular gas reservoir, with significant implications for cloud formation and evolution, the total mass budget and spatial distribution of molecular gas, and the initial conditions of star formation as a related consequence.

astro-ph.GA↗
arXiv2026

Vul4Py: Benchmarking Automated Vulnerability Repair in Python with Paired Exploit and Functional Oracles

Automated Vulnerability Repair (AVR) has advanced rapidly across program analysis, machine learning, and Large Language Models (LLMs), but a verifiable, head-to-head comparison of AVR approaches on Python is still missing. Python underpins critical web, data, and machine-learning infrastructure, yet existing Python benchmarks accept a patch on the strength of a proof-of-concept exploit alone, or apply a functional test only on the subset of entries whose upstream project happens to ship one. Both therefore miss functional regressions, in which a patch defeats the exploit but breaks unrelated behavior. We present Vul4Py, a Python AVR benchmark in which every entry carries a paired oracle: an exploit oracle that must fail on the vulnerable revision and pass on the fixed one, together with a project-native pytest functional oracle that must pass on both. Vul4Py comprises 100 real vulnerabilities from 60 open-source projects, spanning 60 distinct CWEs and the years 2017 to 2025, each packaged with a pinned, reproducible per-instance environment. Using Vul4Py, we compare six approaches in three categories: a specialized vulnerability repair tool, directly prompted LLMs, and software engineering agents. The agents dominate: OpenHands repairs 41 of 100 vulnerabilities, against 4 for the strongest directly prompted LLM and 2 for the specialized tool, despite all three sharing the same backbone model. The paired oracle is what makes these counts trustworthy: it rejects 15 of the 119 patches that an exploit-only oracle would accept, and 98 of the 104 patches it admits are manually confirmed to be semantically equivalent to the developer's patches

cs.SE↗
arXiv2026

D-VLC: Decentralized Vision-Language Collaboration for Heterogeneous Embodied Multi-Robot Systems in Unknown Environments

Multi-robot systems, particularly heterogeneous robot swarms, can improve the efficiency of complex task execution through parallel collaboration and complementary capabilities. However, conventional rule-based methods rely on predefined task models and specialized decision making programs, making it difficult to understand complex semantic instructions and coordinate heterogeneous robots. LLMs introduce strong language understanding and task reasoning capabilities, allowing multi-robot systems to interpret instructions, decompose tasks, and assign roles according to task semantics. VLMs further incorporate visual perception, enabling robots to reason about objects, regions, and spatial relationships in physical environments. Nevertheless, existing LLM/VLM based methods often depend on known maps, centralized and synchronized decision making, limiting their generalization to heterogeneous robots and unseen tasks. We therefore propose a framework that combines decentralized asynchronous reasoning, lightweight information sharing, capability aware collaboration, and a unified action interface, enabling general purpose VLMs to generate robot specific actions executed by learning free experts without task or robot specific training. Experiments across diverse scenarios and multiple VLMs show success rates above 70\%, with completion time reduced by up to 55.8\% relative to the geometric greedy baseline.

cs.RO↗
arXiv2026

ROAD: Reciprocal-Objective Alignment of Discriminative Semantics for 3D Shape Generation

High-fidelity 3D generation predominantly relies on scaling model capacity and data, which incurs prohibitive computational costs. This paradigm typically requires learning geometry from scratch and overlooks the rich semantic and structural priors already encapsulated in discriminative 3D foundation models. We contend that leveraging the profound understanding of the 3D world possessed by these discriminative models can significantly reduce generative cost. To this end, we propose ROAD, a framework that reduces the training cost of 3D generation by transferring these rich discriminative priors into diffusion transformers. To address the inherent semantic-structural heterogeneity between generative and discriminative latents, we introduce a reciprocal-objective alignment strategy. This method synergizes Holistic Semantic Condensing to enforce global semantic coherence and Structural Optimal Alignment, which is formulated as a bipartite matching problem to rigorously align microscopic geometric details between disparate latent spaces. The 3D foundation model is only used for training-time supervision of alignment and is not used at inference, incurring no additional inference cost. Compared with the industrial baseline Step1X-3D, the proposed ROAD achieves highly competitive generation performance with only 1.5% of the training data and significantly reduces training costs, effectively reducing the computational overhead of high-fidelity 3D generation. Code is available at https://github.com/H-EmbodVis/ROAD.

cs.CV↗
arXiv2026

What Resolve Rate Hides: Trajectory Structure Diagnostics for Coding Agents

Coding agents are ranked almost entirely by resolve rate: whether their final patch passes the target tests. Yet two agents can reach the same outcome through very different processes, and a single pass/fail label says nothing about why a run failed or why an accepted run spent extra steps, time, or tokens. This process evidence lives in the trajectory, which records a run's searches, reads, edits, tool calls, validation, and reversions. However, raw traces are heterogeneous and hard to compare across runs. We present TraceProbe, a trajectory-diagnostic framework that recovers what resolve rate hides. TraceProbe normalizes each raw run into a canonical nine-type action taxonomy with deterministic effect labels, then applies two rule-based modules: Insight names single-trajectory anti-patterns adapted from established debugging practice (e.g., search loops, verification skips), while Converge aligns pairs of runs and classifies where their behavior diverges under controlled references. Applying TraceProbe to 2,500 trajectories from five production settings on SWE-Bench Verified, we find that (i) file choice is too coarse to separate success from failure, whereas function selection and completion behavior localize it; (ii) Insight anti-patterns act mainly as corpus-level difficulty clues, with search loops the most stable; and (iii) even resolved runs differ in how quickly they reach relevant code and how much failed work they incur. Trajectory structure thus adds auditable diagnostic context to outcomes by localizing inspection targets, suggesting failure hypotheses, and prioritizing runs for review.

cs.SE↗
arXiv2026

EdgeBench: Unveiling Scaling Laws of Learning from Real-World Environments

Pretraining scaling laws reveal that model capability improves predictably with data and compute. But learning from real world environments after deployment remains far less understood. Analyzing roughly 38,000 hours of agent interaction with the environment across 134 real world tasks, we find, to the best of our knowledge, the first evidence that overall performance during environment learning follows a log-sigmoid scaling law with remarkably high precision, reaching R^2 = 0.998. Across model generations, we also find that agent learning speed roughly doubles every three months. This discovery stems from EdgeBench, a suite of 134 real world tasks with ultra-long horizons, spanning scientific discovery, software engineering, combinatorial optimization, professional knowledge work, formal mathematics, and interactive games. Each task sustains at least 12 hours of continuous agent operation under rich, multilevel feedback, and is built through substantial expert effort. We publicly release 51 tasks and our full evaluation framework to accelerate the study of how agents learn from real world experience.

cs.CL↗
arXiv2026

PRISM: Prioritized Channel Importance with Semi-supervised Domain Adaptation for Cross-Subject EEG Emotion Recognition

Electroencephalogram (EEG) captures endogenous brain activity with high temporal fidelity and holds substantial promise for precise emotion decoding. However, channel redundancy and pronounced inter-subject variability remain key obstacles to scalable generalization. To address these limitations, we propose a novel framework termed PRioritized channel Importance with Semi-supervised doMain adaptation (PRISM), enabling label-efficient cross-subject emotion decoding. On the channel side, PRISM assigns differentiable, data-dependent channel weights via a lightweight expert ensemble, amplifying reliable electrodes while suppressing distractors. On the domain side, PRISM leverages unlabeled data through confidence-filtered pseudo-labels to drive consistency regularization and domain alignment, mitigating subject-specific heterogeneity. Extensive experiments show that PRISM surpasses state-of-the-art methods on DEAP, DREAMER, and SEED datasets, achieving robust cross-subject generalization given limited annotations.

cs.LG↗
arXiv2026

MindAU: EEG-Conditioned Facial Action Unit Editing via Dual-Stream Manifold Alignment

Recent brain decoding studies have made substantial progress in reconstructing externally perceived visual content from neural signals. However, using electroencephalography (EEG) recordings to guide facial expression editing remains largely unexplored and poses a distinct challenge: rather than recovering what a subject sees, it requires identifying facial-action related patterns from noisy EEG signals and grounding them in localized, identity-preserving expression edits. In this paper, we investigate EEG-conditioned facial image editing for fine-grained facial action unit (AU) control and propose MindAU, a unified framework for controlling facial AU edits from EEG signals. MindAU first learns noise-robust and AU-discriminative EEG representations through temporal masked reconstruction and AU classification supervision. It then bridges the modality gap via Dual-Stream Manifold Alignment, aligning EEG features with AU-level text semantics and identity-reduced visual displacement trajectories in the multimodal space of Qwen2.5-VL. Finally, MindAU incorporates EEG-aware Multimodal Rotary Positional Embeddings, landmark-guided reference masking, and AU-aware region supervision into a multimodal diffusion-based editor for high-fidelity identity-preserving editing. We also introduce E-CAFE, a curated benchmark for EEG-Conditioned Action-Unit Facial Editing with paired EEG-face editing samples and standardized evaluation protocols. Extensive experiments demonstrate the effectiveness of MindAU and suggest its potential as a step towards future assistive expression technologies for individuals with facial neuromuscular disorders.

cs.CV↗
arXiv2026

Foundation Model-driven Key Anatomy Frame Selection for Blind-sweep Ultrasound Fetal Birth Weight Estimation

Accurate fetal birth weight (FBW) estimation shortly before delivery is clinically valuable yet challenging due to its reliance on operator expertise, particularly in low-resource settings. To reduce this reliance, we study near-term birth-weight regression from blind-sweep ultrasound (US) videos acquired within 48 hours prior to delivery, with post-delivery weighing as ground truth. Accordingly, we propose a foundation model-driven key anatomy frame selection framework that enables accurate FBW regression despite the absence of plane constraints in blind sweeps. Our highlights are as follows: (1) We believe this is the first work to estimate FBW using blind-sweep US videos, enabling operator-independent assessment. (2) An Anatomy-Guided Frame Selection module equipped with a vision-language foundation model is proposed for keyframe collection in unconstrained sweeps. (3) A Redundancy-Aware Feature Compression module is designed to compress frame features while preserving task-relevant information, alleviating temporal redundancy. Extensively validated on prospectively collected data from 839 patients, our method achieves an MAE of 161.3 g, with 90.23% and 100% of cases falling within 10% and 15% absolute percentage error, outperforming typical Hadlock estimation and strong competitors. Codes are available at https://github.com/ouleoule/BlindSweep-EBW.

cs.CV↗
arXiv2026

Statistical Properties of Molecular Clouds in the Milky Way: Insights from Three-Isotopologue CO Observations of the MWISP Project

We present a comprehensive statistical analysis of molecular cloud (MC) properties using the MWISP survey's 12CO, 13CO, and C18O (J = 1--0) data toward the inner (l = 45$^\circ$--60$^\circ$) and outer (l = 120$^\circ$--130$^\circ$) Galaxy. From a strict selection of 24,724 identified MCs, a final sample of 3,161 well-resolved MCs is established. We investigate the distributions of observational, morphological, and derived physical parameters, as well as their environmental dependencies and intercorrelations. Our analysis reveals that MCs are typically oblate and tend to align with the Galactic disk. A critical evaluation using a nearby subsample confirms significant distance-dependent selection effects for some parameters, nevertheless, the direction of changes in these parameters can indicate distance influence. We also examine several specific subsamples, revealing the distinct characteristics of MCs in the G120 spiral shock region, MCs in the G50 interarm spurs, C18O-bright MCs, and MCs with supra-Larson velocity dispersion. For instance, MCs with supra-Larson velocity dispersion are predominantly small and likely young clouds inheriting turbulence from the diffuse ISM. Notably, a comparison across tracers reveals that typical MCs have a turbulent, diffuse, 12CO-bright gas structure in their outer layers that does not contribute directly to star formation. In contrast, 13CO-bright gas represents a turning point where gravity becomes significant; C18O-bright gas is about gravity-dominated. Comprehensive correlation analysis confirms a flatter $\sigma_v$-size relation than classic Larson's law and a strong mass-size relation. Incorporating dimensional analysis, we derive minimal sets of eigenparameters from which most other observational and physical parameters can be estimated. This highlights the underlying scaling relations that governing cloud properties.

astro-ph.GA↗
arXiv2026

Existence of classical minimal surfaces in $4$ and $5$-manifolds

We prove that every closed Riemannian $4$ or $5$-manifold $M$ contains a branched immersed closed minimal surface. That is, there exists a non-constant weakly conformal harmonic map from some closed Riemann surface into $M$. We rely on the existence of multisections in dimensions $4$ and $5$ to generate a non-trivial class of sweepouts of $M$ by mappings from a closed surface $S$ of genus at least $2$. To each sweepout in a minimizing sequence within the class, through the intermediary of quasiconformal maps of the upper half-plane, we associate a family of hyperbolic metrics on $S$ with respect to which the mappings in the sweepout have nearly equal energy and area. The harmonic replacement method of Colding and Minicozzi is then applied to obtain a min-max sequence that converges to a bubble tree of branched minimal immersions.

math.DG↗