Past ArXiv / September 28, 2026

Daily archive

A preserved paper digest from the previous month.

Archive date 09/28/2026

Source ArXiv

Back to current digest

Reading Queue

ArXiv

9 papers

Benchmark & Evaluation

1 SciHorizon-eLab: An Agentic Protocol-to-Task Compiler for Scalable Benchmarking of Scientific Embodied Agents Maokai Qin, Chuan Qin, Qi Zhang, Dianyu Liu, Zirui Liu, Hongting Niu, Yuanchun Zhou, Hengshu Zhu

Benchmark & Evaluation Embodied Interaction Generative Modeling Retrieval & Grounding

ArXiv

Paper 1 / arXiv:2609.30971 Open arXiv

Relevance 9 Novelty 8

Why selected: Matches criterion 3 very closely: it is an embodied AI paper building a new benchmark and compiler for scientific laboratory tasks, with a novel protocol-to-task generation angle that prior work often ignores.

Embodied agents offer a promising route to automating scientific experimentation, yet their progress is constrained by the lack of reliable and systematic evaluation environments. Existing simulation-based laboratory benchmarks rely heavily on manual task engineering, making it challenging to systematically compile diverse scientific protocols into executable and verifiable embodied tasks at scale. To address this challenge, we introduce SciHorizon-eLab, an agentic protocol-to-task compiler that formulates scientific embodied task construction as a compilation problem. Given a natural-language protocol of scientific experiments, SciHorizon-eLab progressively compiles laboratory protocols into semantic-preserving embodied tasks through semantic grounding, executable task synthesis, and multi-stage simulation-based certification. The system generates semantically grounded environments, executable manipulation programs, and step-level success specifications, while enabling reproducible generation of expert demonstrations and execution traces. Using this pipeline, we further construct \BenchName, a ready-to-use benchmark comprising 300 certified tasks across diverse laboratory operations. It supports HIL task execution, reproducible expert-demonstration generation, and ordered step-level evaluation. Across representative tasks, the strongest policy attains an average success rate of only 49.7%, with further evaluations revealing pronounced weaknesses in human and embodied agent coordination. We publicly release the code, benchmark data, and evaluation toolkit at https://github.com/SciHorizon-elab/SciHorizon-elab.

9 Reliability-Regulated Trajectory Optimization for Progressive COLMAP-Free 3D Gaussian Splatting Zijian Wu, Jinliang Wang, Zidian Lin, Ying Song, Ziqian Lu, Hanjie Ma, Zhen Ye, Mingfeng Jiang

Benchmark & Evaluation LLM Interpretability Time-Series LLM

ArXiv

Paper 9 / arXiv:2609.30865 Open arXiv

Relevance 3 Novelty 6

Why selected: Partial fit to criterion 4 only in a broad sense through 3D reconstruction / view synthesis, but it is not really about vision foundation models or embodied AI benchmarks/methods.

COLMAP-free 3D Gaussian Splatting (3DGS) bypasses computationally expensive structure-from-motion (SfM) pipelines, yet progressive camera pose tracking remains fundamentally vulnerable to error compounding---early pairwise tracking inaccuracies both corrupt subsequent frame initializations and remain permanently frozen in the scene representation. Rather than relying on heavyweight external neural priors or treating progressive tracking through isolated heuristic fixes, we propose a unified reliability-regulated trajectory optimization framework for progressive COLMAP-free 3DGS. At its core, our framework establishes an intrinsic, self-supervised bidirectional cycle-consistency mechanism that systematically regulates progressive camera trajectory estimation across two complementary temporal horizons: (1) Forward Motion Propagation, where the online reliability signal adaptively gates first-order kinematic warm-starts of rigid motion into upcoming pairwise registrations, supplying informed directional search priors while safely intercepting untrusted transitions; and (2) Retrospective Trajectory Correction, where the same reliability signal dynamically weights relative-pose consistency constraints within a sliding window of neighboring camera poses. By governing both prospective state initialization and retrospective trajectory consolidation through a unified reliability regulator, our self-contained framework resolves progressive drift without external priors or offline preprocessing. Extensive evaluations on Tanks and Temples and CO3D-V2 benchmarks show that our method substantially improves camera trajectory accuracy and novel-view rendering quality, outperforming existing unposed baselines. Code is available at https://github.com/Zijian1026/RRTO-CF3DGS.

Vision-Language Reasoning

2 SatNav: A Scalable Benchmark for Long-Horizon UAV Vision-Language Navigation from Satellite Imagery Jiajun Jiang, Chunliang Hua, Zichun Chen, Yanxing Wu, Zeyuan Yang, Jie Song, Xiao Hu

Vision-Language Reasoning Benchmark & Evaluation Embodied Interaction

ArXiv

Paper 2 / arXiv:2609.31507 Open arXiv

Relevance 9 Novelty 7

Why selected: Matches criterion 3 very closely: it is a new embodied AI benchmark for long-horizon UAV vision-language navigation, with scalable simulator-like episode construction and a new method (SwiftVLN) plus satellite-to-UAV transfer.

Urban uncrewed aerial vehicle (UAV) vision-language navigation (VLN) requires agents to follow instructions across extended urban spaces, inherently demanding long-term memory and geospatial grounding. However, scaling existing benchmarks remains difficult because of their reliance on costly reconstructed 3D assets, limiting geographic diversity and episode scale. To address this, we introduce SatNav, a scalable, long-horizon UAV VLN benchmark built from high-resolution satellite imagery. SatNav targets city-level navigation missions and uses satellite crops as approximations of UAV nadir views for visual observations. Through an automated cue-to-episode pipeline, SatNav constructs 118K episodes from 59 scenes across 18 cities, with an average trajectory length of 379 m. To stress-test long-horizon memory and geospatial reasoning, SatNav defines three task families: Boundary, Landmark, and Route, targeting loop progress tracking, landmark-based spatial grounding, and route following with counting cues. Benchmarking classical VLN agents and recent agents based on large vision-language models (LVLMs) on SatNav shows that city-scale navigation remains challenging. We further introduce SwiftVLN, a modular framework with switchable memory components, and conduct systematic memory-design ablations. Finally, satellite-to-UAV transfer experiments show that satellite-trained navigation models can operate on real-flight UAV observations, showing the practical relevance of SatNav. Our project page: https://eku127.github.io/SatNav/

3 DepthEvidence: Unifying Metric Depth Prediction and Geometric Reasoning in Multimodal Language Models Jiangning Wei, Yuan Yao, Miaomiao Cui, Mingsheng Li, Humen Zhong, Shuai Bai, Zhibo Yang

Vision-Language Reasoning Multimodal LLM Benchmark & Evaluation Generative Modeling

ArXiv

Paper 3 / arXiv:2609.31103 Open arXiv

Relevance 8 Novelty 7

Why selected: Matches criterion 1 very closely: it targets spatial reasoning with metric constraints and integrates depth prediction into a multimodal language model for better geometric understanding.

Spatial reasoning with metric constraints requires linking objects to geometric measurements and preserving their numerical content during language reasoning. We present DepthEvidence, a 4B model that uses its own dense metric predictions as object-grounded evidence for language generation. A camera-conditioned decoder predicts full-resolution metric depth using multi-scale visual features and high-resolution RGB refinement. A dense-to-language interface converts predicted depths and decoder features into object-aligned continuous geometry tokens anchored to object identifiers. Geometric supervision encourages metric information to remain recoverable before and after language-context interaction, while instruction tuning supports object measurement and compositional reasoning. We introduce a Depth-VQA benchmark evaluating object-depth queries, relative comparisons, and decisions combining spatial and numerical constraints. Across nine datasets, DepthEvidence achieves the highest average dense $\delta_1$ among evaluated methods, competitive with specialized estimators. It also leads the evaluated methods in instance-level metric depth estimation and overall accuracy on both relative and metric reasoning tracks, while broadly preserving general VQA performance and improving spatial understanding relative to the base model.

4 Diagnosing the Sources of Compositional Failure in Vision-Language Models: A Controlled Analysis Mona Gandhi, Cenk Merih Olcay, Kuan-Chieh Lo, Santiago Castro, Christopher W. Myers, Srinivasan Parthasarathy

Vision-Language Reasoning Benchmark & Evaluation Safety & Backdoors

ArXiv

Paper 4 / arXiv:2609.31456 Open arXiv

Relevance 6 Novelty 7

Why selected: Matches criterion 2 in a broader sense through a controlled analysis of vision-language models, with especially interesting empirical findings about compositional failure rather than a new model.

Vision-language models (VLMs) often struggle with compositional reasoning tasks, but the reasons for this underperformance remain unclear. A common hypothesis is that models struggle to integrate multiple components, leading to training interventions to improve compositional binding. However, this assumption has never been directly quantified. Existing benchmarks evaluate captions only in their composed form, making it impossible to separate the cost of joint reasoning from the cost of recognizing individual components under increasing load. We introduce COMPASS (COMPositional Analysis of SkillS), a controlled evaluation framework designed to isolate and measure the distinct factors underlying compositional failure. By comparing performance on composed captions with their decomposed counterparts , we directly quantify the cost of compositional integration across 87K image-caption pairs. Across multiple VLMs, this gap is real but partial, accounting for only part of the observed degradation. This motivates a finer-grained investigation into what additional factors govern model behavior. We analyze performance at the level of individual skills: object detection, attribute binding, and relation reasoning, using skill-targeted perturbations across 274K image-caption pairs. We find a consistent skill-specific pattern: each skill degrades primarily with the count of its own primitive type (self-load), while cross-load effects are predominantly positive, suggesting that primitives of different types provide useful grounding context. This pattern holds across standard contrastive encoders, explicitly trained compositional reasoning models, and non-contrastive architectures. These findings show that compositional degradation reflects multiple separable factors that cannot be reduced to joint reasoning alone.

5 Open Vocabulary Domain Unlearning Sumanth Udupa, Mehrtash Harandi, Yadan Luo, Mahsa Baktashmotlagh

Vision-Language Reasoning Benchmark & Evaluation Medical AI

ArXiv

Paper 5 / arXiv:2609.31356 Open arXiv

Relevance 5 Novelty 7

Why selected: Matches criterion 4 very closely: it studies VLM behavior and proposes a parameter-editing method for open-vocabulary domain unlearning.

Vision-Language Models (VLMs) exhibit remarkable zero-shot generalization, yet they often encode unwanted or hazardous stylistic domains such as idealized textbook diagrams in medical AI or cartoon vehicles in autonomous driving. Approximate Domain Unlearning (ADU) aims to selectively erase a model's recognition of a target visual domain while preserving accuracy on the remaining domains. However, existing ADU methods operate under a flawed closed-vocabulary assumption: they evaluate unlearning solely on the specific object classes seen during the unlearning fine-tuning phase. Consequently, these methods do not unlearn the domain itself; they merely overfit to seen class-domain pairs, leaving the domain easily recognizable for unseen classes and providing a false sense of removal. We argue that true domain erasure must be class-agnostic. To address this, we formalize Open-Vocabulary Domain Unlearning (OVDU), a rigorous protocol that mandates domain forgetting must transfer to held-out classes. To solve the OVDU challenge, we propose a surgical parameter-editing framework. First, a Fisher Information mask isolates domain-sensitive weights, mathematically protecting foundational zero-shot generalization. Second, our Targeted Manifold Scattering (TMS) objective uses preference-based mining to locally scatter the forget domain's stylistic geometry. Evaluated across PACS, OfficeHome, and DomainNet, our method vastly improves open-vocabulary generalization over existing baselines. Crucially, it delivers exceptional sample efficiency, outperforming peak 8-shot baseline results with only 4 shots.

6 Structure-Guided Masked Autoencoders for Ultra-High Resolution Scientific Image Understanding Enzhi Zhang, Du Wu, Rui Zhong, Cong Ma, Isaac Lyngaas, Amir Koushyar Ziabari, Xiao Wang, Peng Chen, Tao Luo, Toshio Endo, Fumiyoshi Shoji, Kento Sato, Kentaro Uesugi, Takayuki Nonoyama, Ryuji Kiyama, Masahiro Yoshida, Masaru Tezuka, Tetsuya Ishikawa, Satoshi Matsuoka, Masaharu Munetomo, Mohamed Wahib

Vision-Language Reasoning Benchmark & Evaluation LLM Interpretability

ArXiv

Paper 6 / arXiv:2609.30682 Open arXiv

Relevance 5 Novelty 7

Why selected: Matches criterion 4 very closely: it is a vision foundation-model style pretraining method (MAE variant) for ultra-high-resolution scientific images.

Self-supervised pre-training with Vision Transformers, including Masked Autoencoders (MAE), is difficult to apply to gigapixel scientific images. Random masking is poorly matched to the structured, multi-scale morphology of scientific data, while uniform tokenization produces prohibitively long sequences that make $O(N^2)$ attention impractical. We propose SGMA, a structure-guided masked autoencoding framework for ultra-high-resolution scientific images. SGMA couples two components: a content-adaptive quadtree tokenizer that compresses gigapixel images into a fixed-length sequence, and a structure-conditioned masking process that biases reconstruction toward spatially informative regions. To stabilize this process across scales, we introduce Damped Accumulation (DA), which aggregates signal-dependent responses across the tree into a structure canvas used to guide masking. The resulting pre-training task preserves fine microstructure while remaining compatible with standard ViT encoders and MAE-style reconstruction. Across electron microscopy, whole-slide optical microscopy, and X-ray CT datasets, SGMA consistently outperforms MAE baselines. It achieves 95.68% Dice on the 8K x 8K x 28K SpringXCT dataset, improving over the same-architecture MAE baseline by +13.00 points, and 83.21% Dice on the 32K^2 WSI PAIP dataset, improving by +16.84 points, while providing up to a 24.8x inference speedup.

7 Atelier: Learning Local Self-Supervised Features for CryoEM Volumes via Hypernetworks Phillip Lo, Sudarshan Babu, Dari Kimanius, Aly A. Khan

Vision-Language Reasoning

ArXiv

Paper 7 / arXiv:2609.30569 Open arXiv

Relevance 6 Novelty 6

Why selected: Matches criterion 4 reasonably well: it develops a self-supervised feature foundation-style representation for cryoEM volumes and demonstrates its downstream applications in geometry-aware biological analysis.

CryoEM map interpretation requires features that are spatially localized, consistent across samples, and informative across spatial scales. Most deep learning methods for map annotation extract features from fixed voxel grids. However, implicit neural representations (INRs) are able to model volumetric data as scale-agnostic, coordinate-conditioned functions. INRs are therefore attractive for cryoEM, but fitting a separate INR for each map is too expensive for large-scale feature extraction and produces representations that are not aligned across samples. We introduce Atelier, a self-supervised framework that amortizes INR fitting for reconstructed cryoEM maps. Pretrained on 5,439 Electron Microscopy Data Bank maps, Atelier is a transformer-based hypernetwork that generates high-fidelity reconstructions across a wide range of protein structures, including large multi-subunit assemblies. Beyond reconstruction, the INR generated by the pretrained transformer exposes a continuous, local feature field through its intermediate activations at any spatial query point, a property that voxel grid and patch-tokenizer architectures do not naturally provide. Used as auxiliary channels to a 3D nested U-Net annotation head trained from scratch, these coordinate-conditioned features improve performance on eight voxel-level property prediction tasks over a volume-only baseline. Our results demonstrate that amortized implicit neural representations are an effective primitive for geometry-aware analysis of cryoEM data.

8 Pocket-STVG: lightweight architecture for Spatio-Temporal Video Grounding Alberto Presta, Michal Byra, Grzegorz Stefa\'nski, Karol Szurkowski, Eryk Ko{\l}odziejczyk, Krzysztof Arendt

Vision-Language Reasoning Multimodal LLM LLM Interpretability Time-Series LLM

ArXiv

Paper 8 / arXiv:2609.31135 Open arXiv

Relevance 5 Novelty 6

Why selected: Matches criterion 3 very closely: it proposes a new method for spatio-temporal video grounding, which is an embodied/vision-language localization task with an efficiency-focused architecture.

Spatio-Temporal Video Grounding (STVG) aims to localize the spatio-temporal tube in a video corresponding to a natural language query. While recent methods achieve strong performance in fully supervised, weakly supervised, and zero-shot settings, they typically rely on computationally expensive architectures, complex training pipelines, or multimodal large language models. We present Pocket-STVG (P-STVG), a lightweight cascade architecture that addresses STVG by combining efficient pre-trained components instead of large end-to-end models. P-STVG integrates a temporal-aware video encoder based on MobileViCLIP, a spatial encoder-decoder derived from MDETR, and a shared aligned text encoder. Temporal localization is performed through either a lightweight 1D U-Net or a simple thresholding strategy, enabling the same framework to operate in both weakly supervised and zero-shot settings. Furthermore, video representations are precomputed independently of the query, yielding an indexing-friendly pipeline for efficient inference and large-scale video collections. Despite requiring fewer than 90M parameters, P-STVG performs on par with weakly supervised methods and improves on earlier zero-shot approaches at a fraction of their memory and computational cost, establishing a favorable performance-efficiency trade-off for STVG.

Archived papers 9

Top score 17

Average score 13.0