Researchers propose a reliability-aware diffusion planner that distills a vision-language model to generate scene-level reliability heatmaps, guiding UAVs to avoid unreliable regions (e.g., glass, mirrors) in 3D navigation, reducing obstacle violation rate from 40.3% to 9.6% and raising mean reliability from 0.588 to 0.925.

AnyGoal is a training-free multi-robot navigation architecture that uses a Vision-Language Model (VLM) for frontier-based exploration and coordinates agents via a shared 2D Gaussian Bayesian Value Map (BVM), achieving 52.4% subtask success rate on GOAT-Bench, a +27.5pp improvement over Modular GOAT.

ContactWorld benchmark spanning 12 contact-rich manipulation tasks reveals that spatially structured and temporally continuous representations, like point clouds, improve planning success rates from ~20% to 32.1%. Tactile sensing effectiveness depends on cross-modal compatibility; combining point clouds with tactile force fields achieves 36.1%. Tactile information becomes increasingly important for long-horizon planning.

A new study reveals the 'seed lottery' in single-GPU fine-tuning of vision-language-action models: out of 13 identical runs with different seeds, one silently degrades to 65.2% success rate from 91-94%. The culprit is output collapse, where the action predictor produces nearly identical outputs regardless of input. Weight-level regularizers fail to detect this, but output-level regularizers (VICReg, Dropout, or halved learning rate) eliminate all catastrophic seeds. The simplest fix is changing one number in the optimizer config.

This paper proposes an algorithm for efficient domain-adaptive policy learning using kernel representations. It models unknown disturbances with a differentiable random Fourier features kernel approximation. Offline training takes only 50 seconds on an RTX 4090 to optimize the policy via differentiable simulation. During deployment, the policy adapts in real-time by updating kernel parameters through online least-squares estimation. Experiments on Crazyflie quadrotors under various disturbances (wind, ground effects, payload shifts) validate the approach.

This survey examines the shift from isolated vehicle intelligence to multi-agent embodied systems in autonomous driving, focusing on Shared World Models (SWMs) as predictive cross-agent representations. Reviewing over 380 publications, it covers V2X communication, collaborative perception, inter-agent cognition, cooperative planning, end-to-end cooperative driving, and simulation engines. The study finds evaluation concentrated in simulation and offline protocols, with foundation-model-based coordination lacking real-time safety guarantees. Key research priorities include verifiable shared-state maintenance, robust intent and plan alignment, and safe coordinated action under communication and latency constraints.

World models in robot learning predict future states from visual observations and actions. FlowMo-WM infers object-centric motion and long-history context associated with hidden drift without direct flow supervision. It factorizes history into short-history latent and longer-history context, using zero-context residual transition to improve long-horizon prediction in simulated aquatic environments.

This paper presents $\mu_0$, a scalable world model that predicts 3D trajectories of interaction points (objects, tools, hands) rather than pixels or actions, enabling embodiment-agnostic robot learning. The TraceExtract system automatically extracts 3D supervision from videos. Experiments show $\mu_0$ outperforms baselines in trace prediction, and frozen $\mu_0$ can be paired with action experts for downstream tasks, achieving performance competitive with VLA models pretrained with action supervision.

A passive distributed paradigm using deep sub-wavelength acoustic waveguides decouples performance from structural flexibility, achieving high-resolution tactile sensing with sparse microphone arrays.

A novel occupancy-grounded pipeline for hierarchical 3D scene graphs anchors room nodes to tracked free-space regions from occupancy decomposition, providing explicit polygonal footprints. Evaluated on 12 Matterport3D scenes, it recovers more room instances than the state-of-the-art Hydra method but at lower precision, and wall-accurate boundaries remain an open challenge.

Avatar V is a production-scale framework for generating avatar videos that are behaviorally recognizable. It conditions on full token sequences of reference videos, using Sparse Reference Attention, motion representation, and identity-aware super-resolution. Trained on over 100M clips with a five-stage pipeline, it achieves state-of-the-art results on cross-scene benchmarks, outperforming existing systems.

Vision-language models (VLMs) confidently answer image-based questions even without images, inflating benchmark scores without reflecting visual grounding. This study classifies mirages into two regimes: textual biases (relying on language priors) and spurious images (constructing false visual content in latent space). Using a contrastive probing framework and a Prior Harnessing Index (PHI), the authors show these regimes are distinct and require different mitigation strategies—text cleaning addresses only the first, while faithful grounding needs representational interventions.

Researchers propose Temporal Backtracking Search (TBS) to overcome the single-shot bottleneck in generative video reasoning. TBS iteratively generates, verifies, and restarts along the temporal axis, significantly outperforming Best-of-N sampling across algorithmic, navigation, and robotics domains. In out-of-distribution settings, TBS achieves 22.7% success versus 0.7% for BoN.

This paper addresses the lack of explainability in rPPG Transformers by introducing a quantitative evaluation framework. The authors adapt four attribution methods to RhythmFormer's bi-level routing attention, propose a skin coverage metric, and extend the SaCo faithfulness coefficient to rPPG regression. They reveal a multi-hop leakage effect under sparse top-k routing and show that Beyond Intuition mitigates it, achieving the best skin coverage (0.83) and faithfulness (F=0.92) on UBFC-rPPG. The work advances rPPG XAI toward auditable numerical evidence.

Researchers propose i2L (image-to-LoRA), a framework that amortizes style LoRA training into a single forward pass, eliminating per-style optimization. Using an image encoder, learnable LoRA queries, and compressed decoding heads, i2L predicts LoRA weights and outperforms baselines on Z-Image, FLUX.2, and Hidream-O1 in style fidelity, prompt alignment, and perceptual quality.

CineOrchestra is a unified video diffusion model that simultaneously controls subjects, events, cameras, and shot transitions via entity-centric conditioning primitives and two parameter-free coordinated rotary embeddings. It outperforms six per-axis specialists on two new benchmarks.

This paper introduces learnable pairwise connections between filters in CNNs, moving beyond traditional pointwise nonlinearities. The proposed method allows the network to adapt connection functions per layer, improving accuracy.

IoU is a key metric for evaluating spatial alignment between proposals and ground truth, but it has an insensitive region where samples with distinct geometric overlaps yield similar scores. This work introduces morphological similarity metrics (area, shape, aspect ratio) to refine positive sample assignment, producing a supplementary matching score via mean aggregation. Experiments on YOLOv9 show consistent gains on NEUDET and GC10-DET datasets with zero extra inference overhead.

A new study explores how post-training methods like supervised fine-tuning and reinforcement learning can significantly improve generative LLMs for ICD coding, challenging the notion that LLMs are weak medical coders when evaluated solely via prompting.

DLawBench is a new diagnostic benchmark for evaluating LLMs in multi-turn legal consultation. It includes 461 cases from Chinese and U.S. law, categorizing interactions into Cooperative, Dependent, Withdrawn, and Adversarial types. Experiments show the best model, GPT-5.5, achieves only 0.562 on consultation-grounded legal reasoning, exposing sycophancy and a paradox where models perform worse when clients need guidance most.

This paper presents SANA (Search Agent Navigation Ablation framework), a diagnostic ablation framework that decomposes failures in exploratory question answering (EQA) over data lakes. By transforming EQA tasks into runtime profiles with gold source sequence, sanitized subquestions, and execution records, SANA constructs idealized search, planning, and data-analysis tools for component ablation, with residual gaps indicating policy failures. Experiments on LakeQA and KramaBench show data analysis as a consistent bottleneck, search as a major limitation in large data lakes, and planning as less critical.

Neural topic models enable scalable semantic discovery, but their integration with quantum hardware remains largely unexplored. This paper presents a proof-of-concept hybrid classical-quantum variational autoencoder (VAE) for topic modeling, embedding parameterized quantum circuits within the VAE inference network while retaining a classical topic-word decoder. To address quantum hardware resource constraints, a modified Gaussian Softmax posterior is proposed that decouples latent space dimensionality from the number of topics, enabling operation on a 10-qubit quantum device. On the AgNews dataset, the hybrid VAE outperforms state-of-the-art neural topic models, achieving a C_v coherence score of 0.71 and NPMI of 0.20 while preserving high topic diversity. A fully classical variant also o…

This paper introduces a validation framework to evaluate whether LLM-based urban simulators produce empirically realistic human mobility patterns or merely plausible narratives. Testing AgentSociety and CitySim on datasets from Paris and Shanghai reveals a substantial gap between narrative plausibility and empirical realism, especially in spatial and temporal constraints. The authors also provide open-source tools for scalable evaluation.

Current cultural alignment approaches focus on inference-time interventions, assuming models already contain sufficient cultural knowledge. This paper argues modern LLM pipelines suffer from a 'cultural data funnel,' where explicit cultural signals decline sharply during post-training, while geographically concentrated, task-specialized data dominates. Multilinguality enhances geographic diversity but does not ensure balanced representation. The authors release a culturally tagged dataset with 5.6M samples to facilitate further research.

This paper overviews the QIAS 2026 shared task on Islamic inheritance reasoning, using the MAWARITH benchmark of 12,500 Arabic cases. 16 teams participated, and results show LLMs struggle with precise legal and numerical reasoning.

This paper presents the participation of team PSL in the QIAS 2026 Shared Task on Arabic Islamic inheritance reasoning. The task evaluates large language models on solving inheritance cases requiring legal interpretation, multi-step reasoning, and precise numerical computation. Results show commercial models like Gemini 2.5 Flash outperform open-source models in identifying heirs, applying exclusion rules, and maintaining consistency, while open-source models exhibit instability.

A new study introduces WebDecept, a framework for injecting deceptive interfaces to test web agent safety. Evaluations show current agents are highly susceptible to manipulations like targeted ads and domain redirection, and prompt-based defenses often fail.