Benchmarking and Evolving Reason-Reflect-Rectify for Reflective Visual Generation

Text-to-Image (T2I) models and Unified Multimodal Models (UMMs) have achieved remarkable progress in visual generation. However, their reliance on a single-pass generation paradigm limits their ability to handle complex prompts requiring iterative refinement. To enable multi-round Reflective Visual Generation (RVG), we formalize the Reason-Reflect-Rectify (R^3) loop as a core framework and introduce R^3-Bench, a benchmark of over 600 expert-annotated instances that quantifies iterative reasoning and rectification capabilities. Evaluation on R^3-Bench reveals a critical gap: while state-of-the-art models can identify generation errors, they fail to generate actionable rectification instructions. To bridge this gap, we propose R^3-Refiner, a dual-stage framework leveraging Group Relative Policy Optimization (GRPO) and a Hierarchical Reward Mechanism (HRM) to better align rectification with reflective reasoning. Experiments show that R^3-Refiner achieves significant improvements on R^3-Bench (+12.0% in Reflective Verdict Score, +9.0% in Rectification Score), and can be seamlessly integrated with various MLLMs to enhance the generation quality of different T2I models on GenEval++ and T2I-CompBench. Code is available at https://github.com/xiaomoguhz/R3-Bench.

Sparse Fluid Antenna Arrays: Continuous Position Design Beyond Classical DOF Limits

Fluid antenna system (FAS), which continuously repositions a single physical element across a deployment region $[0, D]$, breaks this limit by freeing antenna positions from the discrete grid entirely. This paper establishes the theoretical foundations of sparse FAS design for direction-of-arrival (DOA) estimation and shows that continuous position freedom unlocks three compounding advantages over the classical designs. \emph{First}, we derive a universal dual DOF bound and prove that FAS-optimized positions can approach it, growing the DOF linearly with $D/λ$ , where $λ$ is the signal wavelength, rather than saturating at $O(N^2)$. \emph{Second}, the CRB scales as $O(1/D^{2L})$ for $L$ sources, a $(D/(N^2 d_0))^{2L}$ improvement over the best grid design, with $d_0 = λ/2$ and D-optimal positions admitting closed-form solution for single sources and efficient Frank-Wolfe algorithm for multiple sources. \emph{Third}, we propose a two-stage FAS-MUSIC approach that combines coarray MUSIC disambiguation with full-aperture local maximum likelihood (ML) refinement to track the CRB, overcoming the grating-lobe ambiguity inherent in large-aperture non-uniform arrays. Robustness to minimum spacing constraints, mutual coupling, and finite position accuracy is also analyzed. Extensive simulations show that FAS-MUSIC achieves $17.5\times$ lower root mean squared error (RMSE) than uniform linear array (ULA) MUSIC and that FAS with $4$ antennas outperforms MRA with $8$ antennas, gains that are unattainable by any grid-constrained design.

Large Vision-Language Models Get Lost in Attention

Despite the rapid evolution of training paradigms, the decoder backbone of large vision--language models (LVLMs) remains fundamentally rooted in the residual-connection Transformer architecture. Therefore, deciphering the distinct roles of internal modules is critical for understanding model mechanics and guiding architectural optimization. While prior statistical approaches have provided valuable attribution-based insights, they often lack a unified theoretical basis. To bridge this gap, we propose a unified framework grounded in information theory and geometry to quantify the geometric and entropic nature of residual updates. Applying this unified framework reveals a fundamental functional decoupling: Attention acts as a subspace-preserving operator focused on reconfiguration, whereas FFNs serve as subspace-expanding operators driving semantic innovation. Strikingly, further experiments demonstrate that replacing learned attention weights with predefined values (e.g., Gaussian noise) yields comparable or even superior performance across a majority of datasets relative to vanilla models. These results expose severe misallocation and redundancy in current mechanisms, suggesting that state-of-the-art LVLMs effectively ``get lost in attention'' rather than efficiently leveraging visual context.

Large Language Model based Interactive Decision-Making for Autonomous Driving

In high-conflict mixed-traffic scenarios involving human-driven and autonomous vehicles, most existing autonomous driving systems default to overly conservative behaviors, lack proactive interaction, and consequently suffer from limited public acceptance. To mitigate intent misunderstandings and decision failures, we present a Large Language Model based interactive decision-making framework that augments scene understanding and intent-aware interaction to jointly improve safety and efficiency. The approach uses Object-Process Methodology to semantically model complex multi-vehicle scenes, abstracting low-level perceptual data into objects, processes, and relations, thereby streamlining reasoning over latent causal structure. Building on this representation, the Large Language Model parses both explicit and implicit intents of surrounding agents and, under jointly enforced safety and efficiency constraints, selects candidate maneuvers. We further generate perturbed trajectory candidates via Monte Carlo sampling and evaluate them to obtain an optimized executable trajectory. To foster transparency and coordination with nearby road users, the final decision is translated by the Large Language Model into concise natural-language messages and broadcast through an external Human-Machine Interface, completing a closed loop from scene understanding to action to language. Experiments in a cluster driving simulator demonstrate that the proposed method outperforms traditional baselines across safety, comfort, and efficiency metrics, while a Turing-test-style evaluation indicates a high degree of human-likeness in decision making. Besides, these results suggest that coupling semantic scene abstraction with Large Language Model mediated intent reasoning and language-based eHMI communication offers a practical pathway toward interactive, trustworthy autonomous driving in dense mixed traffic.

OVPD: A Virtual-Physical Fusion Testing Dataset of OnSite Auton-omous Driving Challenge

The rapid iteration of autonomous driving algorithms has created a growing demand for high-fidelity, replayable, and diagnosable testing data. However, many public datasets lack real vehicle dynamics feedback and closed-loop interaction with surrounding traffic and road infrastructure, limiting their ability to reflect deployment readiness. To address this gap, we present OVPD (OnSite Virtual-Physical Dataset), a virtual-physical fusion testing dataset released from the 2025 OnSite Autonomous Driving Challenge. Centered on real-vehicle-in-the-loop testing, OVPD integrates virtual background traffic with vehicle-infrastructure perception to build controllable and interactive closed-loop test environments on a proving ground. The dataset contains 20 testing clips from 20 teams over a scenario chain of 15 atomic scenarios, totaling nearly 3 hours of multi-modal data, including vehicle trajectories and states, control commands, and digital-twin-rendered surround-view observations. OVPD supports long-tail planning and decision-making validation, open-loop or platform-enabled closed-loop evaluation, and comprehensive assessment across safety, efficiency, comfort, rule compliance, and traffic impact, providing actionable evidence for failure diagnosis and iterative improvement. The dataset is available via: https://huggingface.co/datasets/Yuhang253820/Onsite_OPVD

Hybrid Architecture Gets Fluid: A New Paradigm for Direction-of-arrival Estimation in 6G Networks

High-precision direction-of-arrival (DOA) estimation, as a key sensing capability for 6G-enabled applications such as autonomous driving and extended reality, is increasingly dependent on the effective exploitation of spatial degrees of freedom (DOFs). This paper integrates two frontier DOFs-oriented paradigms and proposes a fluid antenna-enabled hybrid analog-digital (FA-HAD) architecture, which features an extremely lightweight front-end configuration mechanism and efficient spatial DOFs exploitation. Within this architecture, a collaborative spatial-phase sampling strategy is first developed to enable real-time 2-D DOA estimation under compressive observations, and a single-source CRLB analysis is provided to quantify the achievable performance limit, offering quantitative guidance for accuracy-overhead trade-offs. Furthermore, an efficient virtual-array spatial covariance matrix reconstruction method is proposed to recover a physically meaningful covariance representation, thereby providing a covariance-domain interface that is directly reusable by a broad class of existing covariance-based array processing and array design techniques, which strengthens the scalability and transferability of the proposed architecture. Building upon the reconstructed SCM, a Jacobi-Anger expansion based dimension-reduced MUSIC estimator is further derived for arbitrary planar arrays with a favorable computational cost. Simulation results demonstrate that the proposed FA-HAD framework attains DOA accuracy close to fully digital systems while substantially reducing RF hardware complexity and training overhead.

HyperLiDAR: Adaptive Post-Deployment LiDAR Segmentation via Hyperdimensional Computing

LiDAR semantic segmentation plays a pivotal role in 3D scene understanding for edge applications such as autonomous driving. However, significant challenges remain for real-world deployments, particularly for on-device post-deployment adaptation. Real-world environments can shift as the system navigates through different locations, leading to substantial performance degradation without effective and timely model adaptation. Furthermore, edge systems operate under strict computational and energy constraints, making it infeasible to adapt conventional segmentation models (based on large neural networks) directly on-device. To address the above challenges, we introduce HyperLiDAR, the first lightweight, post-deployment LiDAR segmentation framework based on Hyperdimensional Computing (HDC). The design of HyperLiDAR fully leverages the fast learning and high efficiency of HDC, inspired by how the human brain processes information. To further improve the adaptation efficiency, we identify the high data volume per scan as a key bottleneck and introduce a buffer selection strategy that focuses learning on the most informative points. We conduct extensive evaluations on two state-of-the-art LiDAR segmentation benchmarks and two representative devices. Our results show that HyperLiDAR outperforms or achieves comparable adaptation performance to state-of-the-art segmentation methods, while achieving up to a 13.8x speedup in retraining.

Fundamental Analysis of Scalable Fluid Antenna Systems: Identifiability Limits, Information Theory, and Joint Processing

Unlike fixed-position arrays with static observation entropy, the scalable fluid antenna system (S-FAS) can dynamically adjust its aperture to form different observation spaces with configuration-dependent entropy budgets. This reconfigurability requires an information-theoretic framework beyond traditional algebraic identifiability analysis. This paper establishes an observation entropy framework for S-FAS, which unifies the derivation of identifiability limits, the diagnosis of processing bottlenecks, and system design optimization. For an S-FAS with mutual coupling suppression, we derive a complete capacity hierarchy among compressed, extended, and jointly stacked configurations. The entropy framework reveals that sequential two-stage processing suffers from an information bottleneck that restricts achievable capacity, while the noise entropy ratio can be used to distinguish fundamental performance limits from algorithmic deficiencies. A joint MUSIC algorithm is proposed to approach the theoretical joint capacity bound. Extensive Monte Carlo simulations, validated by both algebraic and information-theoretic criteria, verify the derived capacity hierarchy and identifiability boundaries.

DOA Estimation for Low-Altitude Networks: HAD Architectures, Methods, and Challenges

With the rapid expansion of low-altitude economy (LAE) services and the growing demand for integrated sensing and communication (ISAC) in air-ground networks, reliable direction-of-arrival (DOA) estimation has become essential for both directional communication and sensing functions. DOA underpins beam alignment, spatial-reuse scheduling, and ISAC-critical tasks such as airspace situational awareness and multi-target monitoring. Hybrid analog-digital (HAD) architectures have emerged as a practical solution for large-aperture directional operation under stringent radio frequency (RF), analog-to-digital converter (ADC), and size, weight, and power (SWaP) constraints. However, HAD compresses antenna-domain observations through analog combining, fundamentally reshaping the measurement model and introducing new algorithmic and system-level challenges for DOA estimation. This article first reviews the principles and representative architectures of HAD, highlighting their advantages for scalable beam-centric and ISAC-oriented operation in LAE scenarios. We then provide a structured overview of HAD-enabled DOA estimation methodologies, including spatial covariance matrix (SCM) reconstruction, multi-combiner scan-based acquisition, and pilot-aided estimation, along with key design tradeoffs. Finally, we discuss open challenges and outline reliability-driven research directions toward robust, deployable HAD-enabled DOA solutions for practical ISAC-enabled low-altitude environments.

KLDrive: Fine-Grained 3D Scene Reasoning for Autonomous Driving based on Knowledge Graph

Autonomous driving requires reliable reasoning over fine-grained 3D scene facts. Fine-grained question answering over multi-modal driving observations provides a natural way to evaluate this capability, yet existing perception pipelines and driving-oriented large language model (LLM) methods still suffer from unreliable scene facts, hallucinations, opaque reasoning, and heavy reliance on task-specific training. We present KLDrive, the first knowledge-graph-augmented LLM reasoning framework for fine-grained question answering in autonomous driving. KLDrive addresses this problem through designing two tightly coupled components: an energy-based scene fact construction module that consolidates multi-source evidence into a reliable scene knowledge graph, and an LLM agent that performs fact-grounded reasoning over a constrained action space under explicit structural constraints. By combining structured prompting with few-shot in-context exemplars, the framework adapts to diverse reasoning tasks without heavy task-specific fine-tuning. Experiments on two large-scale autonomous-driving QA benchmarks show that KLDrive outperforms prior state-of-the-art methods, achieving the best overall accuracy of 65.04% on NuScenes-QA and the best SPICE score of 42.45 on GVQA. On counting, the most challenging factual reasoning task, it improves over the strongest baseline by 46.01 percentage points, demonstrating substantially reduced hallucinations and the benefit of coupling reliable scene fact construction with explicit reasoning.