Observation

Agent trajectories record what an agent does and what happens next. Yet standard supervised fine-tuning (SFT) applies loss only to agent-authored action tokens, using environment observations as context but not as prediction targets. We ask whether this convention provides the best initialization for subsequent reinforcement learning. We introduce ActObs, which also supervises the observation tokens already present in each trajectory. Although deployed agents never generate observations, learning to predict them encourages the policy to model action consequences without adding data, parameters, sequence tokens, or forward passes. The methods perform similarly after SFT but diverge after GRPO. On Qwen3-4B, GRPO from ActObs achieves higher pass@k at every evaluated sampling budget than its action-only counterpart on Terminal-Bench 2.0. On Qwen3-8B, it trades some pass@1 reliability for higher pass@k (+3.4 pp at pass@16) and solves more distinct tasks. The advantage extends to cross-domain code editing on aider-polyglot (+4.2 pp at pass@1 at 4B), whose tasks are unseen during SFT and RL. ActObs retains more entropy during RL while requiring less policy movement, leaving the final policy closer to its SFT initialization. Our analysis traces this difference to SFT: action and observation gradients rapidly become orthogonal, while action-only training leaves a large residual observation gradient and degrades environment prediction below the base model. Joint supervision prevents this one-sided specialization, preserving consequence prediction and preparing the policy for downstream exploration.

Visuomotor policies for multi-fingered dexterous manipulation are highly sensitive to camera viewpoint shifts. To achieve view invariance, recent methods increasingly rely on explicit 3D modalities like RGB-D or point clouds, which can introduce hardware dependencies, calibration requirements, and vulnerability to sensor noise during real-world deployment. In this work, we show that view-invariant control can be achieved without explicit test-time 3D sensing by encoding geometric knowledge into the visual representation during simulation. We present AnyViewDex, an asymmetric training pipeline that combines multi-view contrastive alignment with privileged 3D geometric supervision. By regressing absolute 3D object coordinates during simulated training, this auxiliary objective provides a geometric grounding signal that mitigates the spatial collapse of the globally pooled contrastive embedding. At deployment, the policy operates zero-shot using only uncalibrated monocular RGB and proprioception. We validate this approach across both reinforcement learning and student-teacher distillation. In hardware evaluation on an xArm7 with a 16-DoF LEAP Hand, AnyViewDex reaches 76.7% grasping success across eight unseen objects and six uncalibrated viewpoints (480 trials; 2,400 across all ablation conditions), indicating that geometrically grounded monocular policies transfer zero-shot without test-time depth. Project Page: https://anyviewdex.github.io/

A low-impedance robot finger reports through joint torque how strongly it is loaded, but the same torque can arise from a small force near the fingertip or a large force near the joint. Resolving the force therefore requires knowing where along the finger contact occurred. LUMO makes that location externally observable. Embedded LEDs illuminate a compliant silicone pad, and contact deforms the pad so that light emerging from the finger's side changes in a pattern set by where the load acts. Because the same structure also carries the contact load, we optimize its cross-section, including the pad profile, rigid carrier, and lateral void, for two behaviors at once. Mechanically, the pad conforms under low preload while the carrier increasingly restricts further deformation as load rises. Optically, different contact locations produce separated responses on the finger's side. The search uses rigid--soft contact simulation, ray tracing, and multi-objective Bayesian optimization. Across two silicones, six contact locations, and 10- and 30-mm spherical indenters, the optimized morphologies improve neighboring-location separation relative to variation from re-establishing contact by $15$--$59\%$. Estimating contact location from the optical response using the known LED spacing and combining it with joint torque gives $1.44~\mathrm{N}$ normal-force MAE over 931 samples. In a two-finger hand, localized side responses appear on several links simultaneously during grasps.

Real-world Earth observation (EO) agents must translate high-level scientific questions into executable workflows to acquire observations, prepare data, perform domain computations, and derive conclusions from runtime evidence. Existing EO agents typically start from supplied observations, while benchmarks typically provide prepared inputs or candidate answers, leaving full-chain open-world EO execution largely untested. We present Earth-Agent-Pro, an execution-adaptive Plan-and-Execute framework using expert-authored skills to constrain planning and runtime tool use. Workflow-centered structured memory records planned steps, accepted evidence, and their dependencies, enabling repair of only the affected workflow suffix when runtime evidence invalidates a step. Separate large language model adapters use sequence-level supervised fine-tuning for planner workflow composition and node-level group relative policy optimization with locally verifiable rewards for executor tool-argument grounding. Earth-Bench-Pro instantiates 248 expert-curated task cores as 744 questions under three matched regimes. Its 248 Open-World Execution questions span RGB imagery, spectral observations, and remote sensing products, pairing high-level requests with runtime data requirements, executable trajectories, and open-ended answers grounded in execution evidence. With a shared GPT-5 backbone, Earth-Agent-Pro achieves 66.13% LLM-as-Judge accuracy, exceeding ReAct by 20.95 points in this metric and 24.44 points in Tools-In-Order. Joint adapter tuning raises Qwen3.5-9B LLM-as-Judge accuracy from 38.31% to 50.00%, an 11.69-point gain over the untuned configuration. Planning-only evaluation and execution with the reference workflow show that the adapters improve workflow composition and argument grounding, respectively. Code and datasets will be released soon.

Methane plume detection from satellite imagery is constrained by incomplete observations: public satellites provide complementary spatial, spectral, and atmospheric evidence, but real plume cases rarely contain fully paired multi-sensor measurements because of revisit schedules, cloud coverage, acquisition quality, and the transient nature of emissions. Most learning-based detectors rely on single-sensor inputs, especially Sentinel-2 (S2), leaving many reported plume cases unusable. We construct MethaneUnion, a temporal multi-sensor dataset built from Carbon Mapper plume reports and matched S2, Landsat 8/9 (L8/9), EMIT, and Sentinel-5P (S5P) observations. Built on MethaneUnion, MethaneFuse learns from heterogeneous satellite observations under partial sensor availability without requiring complete four-sensor measurements. MethaneUnion expands usable coverage from 3,211 valid S2-matched plume cases to 8,981 reported plume cases with multi-sensor observations. At the representative 480 m setting, MethaneFuse achieves 84.87 F1 and 93.62 AUROC, improving over the strongest baseline by 5.65 F1 and 8.30 AUROC points while reducing false positives by 8.19 points. Sensor-availability experiments show that MethaneFuse improves detection when S2 is available and transfers plume knowledge to L8/9, EMIT, and S5P when S2 is unavailable. These results demonstrate the value of learning from incomplete heterogeneous sensor observations for practical methane plume detection.

We present DCLP++, a local navigation frameworkthat uses footprint clearance as the geometric basis for studying relative motion features in dynamic environments. Each valid LiDAR return is mapped to its shortest Euclidean distance from the filled robot footprint before reciprocal encoding, replacing distance from the sensor with distance to the occupied body. Radial measurementsor simulated planar relative velocities provide short-horizon features without static-dynamic labels in the policy input. A preliminary study uses a rectangular robot with a speed limit of 1 m/s among 20 moving obstacles. On 100 fixed validation tasks, two selected training seeds yield mean success rates of 42% with sensor rangeand 70% with footprint clearance after 200,000 environment steps.Motion variants show mixed additional gains. These results supportthe clearance-based observation in the evaluated setting; reliable motion benefits and transfer across robots require further evaluation.

We present GOLF, the first-place solution to the SHOW3D Interaction Field Estimation Challenge at HANDS@ECCV 2026. Given synchronized egocentric stereo views, the task is to predict a 3D vector from each of 21 hand joints to the closest point on the manipulated object. GOLF combines dense global context, locally sampled hand/object evidence, and common-frame Plücker-ray geometry. We adapt DINOv3 ViT-H+/16 with LoRA and trainable LayerNorm parameters, then jointly decode both interaction fields. Our primary model achieves an official score of 27.61 and a mean ADE of 27.96 mm on the hidden test set. An equal-weight ensemble with a complementary directly fine-tuned variant improves these results to an official score of 27.47 and a mean ADE of 27.82 mm, securing first place.

Learning from Observation (LfO) is a fundamental robotic capability that replicates how humans and animals socially learn from each other. Beyond its biological parallels, this modality provides a practical solution for data scaling in sample-inefficient and data-starved domains like robotics. Recent work has demonstrated promising results in learning manipulation skills from human videos, yet progress in this area remains difficult to assess. Existing methods vary widely in assumptions, hardware choices, and environment setups making it difficult to draw meaningful comparisons and identify advances in the field. To address these challenges, we introduce RoboReel: a unified benchmark for evaluating models that learn policies from human videos. RoboReel consists of bundled real-world human demonstration videos, simulated robot trajectories, and evaluation environments on ten manipulation tasks. We develop four test suites to evaluate the models' performance on multiple axes, including the robustness to visual distractors and the ability to complete long-horizon tasks. Our benchmark covers learning-from-observation models from different categories, and studies the effectiveness of multiple representation choices in our benchmark evaluation that covers over seven state-of-the-art algorithms (including our VLA based variants) in the field of LfO. Finally, we present an analysis of the different types of algorithms showing that long-horizon tasks and tasks with low tolerances are still challenging for current models. Webpage: https://roboreel.github.io

Driver alerting from dashcam video requires sequential decision-making under partial observability: a system must decide not only whether a scene is risky, but also when the evidence is sufficient to warn. Most existing accident anticipation models output a binary risk score, leaving ambiguous scenes to be handled by thresholding. We propose VLAlert, a vision-language alerting framework that casts warning generation as a tri-action policy over SILENT, OBSERVE, and ALERT. The OBSERVE action acts as an internal evidence-gathering decision that delays uncertain warnings and changes the next observation window, creating a lightweight perception-action loop for adaptive alerting. VLAlert uses Qwen3-VL-4B as a safety-evidence generator and pools hidden states from structured belief spans to form compact representations for danger estimation and policy prediction. We evaluate VLAlert on VLAlert-Bench, a unified per-tick benchmark from four real-world dashcam alert datasets, and further test transfer to held-out naturalistic ADAS takeover clips. On VLAlert-Bench validation, VLAlert achieves the highest deployment-oriented utility among tested baselines, with DAUS 0.4878 compared with 0.4752 for Open-BADAS, and improves AUROC, AP_tick, F1_t, and balanced accuracy from 0.610, 0.176, 0.276, and 0.581 to 0.689, 0.195, 0.297, and 0.648, respectively. On 221 held-out ADAS-TO-Critic clips, VLAlert improves R@5s from 74.2% to 88.7% and F1 from 0.585 to 0.686. These results indicate that adaptive observation and safety-focused VLM representations provide measurable gains for driver-facing alert decisions.

Learned world models typically assume that observations arrive synchronously, an abstraction inherited from simulators that return a complete state vector at each environment step. Physical sensing instead operates at heterogeneous rates, leaving most observation channels stale at any given instant. Interpolating stale channels introduces measurements that were never observed, while downsampling to the slowest sensor discards valid measurements. A natural alternative is to zero-order-hold the most recent reading and provide the known sampling schedule to the model through two features, staleness and time-to-refresh. We test this prediction using transformer world models across three regimes of increasing causal coupling: open-loop rollouts in continuous-control locomotion, closed-loop model-predictive planning in which each learned model serves as the planner dynamics, and a linear latched-actuator system in which refresh events apply a zero-order-held command to the plant. Our findings show that the effectiveness of time-to-refresh depends on the causal role of the sampling schedule, specifically when refresh events affect the system rather than merely report its state. These results establish when sampling schedules provide useful information for predictive world models operating under asynchronous physical observations.

Learning-by-Demonstration (LbD) enables intuitive robot programming by capturing expert skills, which is crucial for agility in high-mix, low- volume manufacturing. This systematic literature review analyzes passive LbD for industrial assembly processes, focusing on the perception architecture and the generalization of the perceived demonstration. We specifically investigate one-shot approaches where only a single demonstration is required. The review evaluates how systems adapt to new assemblies using this limited data. We identify a shift towards object-centric perception, allowing learned primitives to be transferred to new product variants with minimal training.

Pass receiver selection is a fundamental task in football analytics, aiming to predict the intended receiver under a given game state. This task is challenging with event-centered freeze-frame observations, a broadcast-like setting that provides only partial and variable player visibility without complete trajectories or stable player identities. The model must therefore reason over anonymous visible candidates, opponent pressure, and recent context under partial observation. To address this setting, we propose a Hierarchical Possession-aware Graph Pointer Network (HPGPN), which formulates pass receiver selection as variable-size candidate prediction over visible teammates. HPGPN jointly models current player interactions, local event context, and possession-level temporal dynamics. It represents the current pass situation with a graph, incorporates fixed event context, and uses dynamic possession history to capture how the attacking sequence evolves. Candidate representations are refined hierarchically by integrating spatial, contextual, and historical evidence, and a glimpse pointer head scores the receiver candidates. Experiments on public football event and freeze-frame data show that HPGPN improves pass receiver selection performance. Ablation studies demonstrate the effectiveness of graph-based interaction modeling, fixed event context, and dual-branch dynamic possession-history modeling.

Recent advances in Earth Observation representation learning accommodate heterogeneous sensors and missing observations, often through larger architectures. We present MEOX (Multimodal Earth Observation with eXperts), a multimodal masked autoencoder with a 2.939 million-parameter encoder and 3.115 million parameters in total. Sensor-specific adapters, explicit validity signals, and a shared sparse-expert block preserve modality-dependent processing before a learned patch-wise fusion. Four metadata tokens then accompany a single spatial sequence through fourteen further encoder blocks. Shared expert projections with private low-rank residuals constrain parameter growth, while rotary attention supports downstream spatial grids different from pretraining. The model is pretrained on 1.228 million MMEarth64 samples using modality-balanced masked reconstruction and structured sensor dropout. Frozen transfer is evaluated on six GEO-Bench tasks at both 64 and 224 pixels. The model reaches 64.42% mean intersection-over-union on cashew segmentation at 64 pixels and 90.56% average accuracy on EuroSAT at 224 pixels, exceeding the corresponding reported CSMoE results. BigEarthNet finetuning reaches 72.95% micro-average precision. Routing diagnostics distinguish expert participation, spatial dependence, modality association, and functional contribution. A held-out WorldCover probe measures a 0.64-percentage-point benefit from metadata, while retrieval separates same-sensor semantics from cross-sensor alignment. These results demonstrate sensor-flexible representation learning and strong task transfer using a compact parameter budget.

State-of-the-art AI weather models have shown impressive medium-range forecast skill and computational efficiency, but suffer two key shortcomings: their forecasts have lower spatial and temporal resolution than the best physics-based models and they are exclusively initialized with and trained on analysis data. As a result, they cannot directly make use of observations, and any biases in the analysis are inherited by the forecast. WeatherNext 3 addresses these shortcomings and establishes a new state-of-the-art for probabilistic medium-range forecasting skill. First, WeatherNext 3 generates new forecasts every hour (rather than every 6 hours like traditional global models) by ingesting low-latency geostationary satellite data. Second, WeatherNext 3's temporal and spatial resolution are on par with physics-based global models, with hourly time steps and 0.1 degree resolution for single-level variables, including solar radiation and cloud cover. Third, WeatherNext 3 moves beyond traditional analysis variables by learning to predict satellite-derived precipitation estimates, as well as tropical cyclone and station observations. Modelling sparse station data allows WeatherNext 3 to make 2m temperature and dewpoint predictions at any location and time, conditioned on local geographical features, with substantially lower error than competing global models, even when evaluated against unseen stations. Together, WeatherNext 3's capabilities move operational AI-based weather forecasting beyond emulating the traditionally distinct stages of data assimilation, forecasting and post-processing, which helps to further push the frontier of performance and granularity for global weather prediction.

Mechanistic interpretability is increasingly used to guide interventions such as activation steering, circuit removal, and safety monitoring. Yet an internal estimate that is accurate on average can still choose a poor action. We present ObserverBench, a benchmark framework for testing whether an internal estimator---an observer---is adequate for the intervention, control, or safety task it directs. Each task fixes the model, information boundary, allowed actions, decision rule, held-out cases, and loss. The benchmark reports estimation accuracy separately from the loss caused by the chosen action. Theory and experiments show why both are needed. In closed-loop control, observer errors matter at the starting point and along directions the allowed intervention can reach. On circuit-intervention tasks in GPT-2-small and Qwen2.5-7B, pairwise observers predict unseen effects more accurately without always choosing better actions; observers trained on action loss choose lower-loss actions. In safety triage, a score that perfectly separates violations can allocate a fixed intervention budget poorly when violations have different costs. Across Qwen2.5-7B, Gemma-2-9B-it, and prospectively frozen Qwen3.5-9B APPS tasks, AUROC can rank monitors differently from deployment loss, and the best information source changes across models. Sparse SAE readouts also trail their layer-matched dense controls on the reported Qwen panels, under disclosed activation-density or checkpoint mismatches. ObserverBench provides fixed task contracts, runnable baselines, and table-based submissions for evaluating interpretability methods through the actions they enable.

A long-horizon agent's trace outgrows both of its consumers: the human observer monitoring the run, and the agent itself, whose bounded context the trace must be folded back into. We present a live trace model, an append-only event ledger folded incrementally into typed run state and compiled into per-consumer views, and evaluate it for both consumers against deterministic ground truth. For the observer side, evaluated with an LLM reader as proxy, the compiled view answers monitoring questions using approximately 14x and 15x fewer input tokens (by reader) and at 5-7x lower cost than a budget-capped single-call reading of the raw trace, with higher accuracy (0.85-0.87 versus 0.48). Because the questions were co-designed with the view schema, we treat the token and cost reduction, conditional on schema coverage, as the transferable result. For the agent, on 120-link sequential-dependency tasks, mechanisms that maintain the task's running statistic in per-step state succeed where full-context prompting fails (30/30 versus 8/30 under a clean protocol, n=30, labeled descriptive owing to benchmark-system co-development); a prompt-level scratchpad matches the fold's accuracy at lower cost, and a two-arm decomposition attributes the fold's accuracy to its deterministic aggregate and its cost advantage to its compactness. The fold's remaining value over cheaper alternatives is deterministic auditability and serving the observer from the same state. We derive eleven candidate requirements for trace folding from observed failures and delimit them with an order-sensitive task family on which the fold ceases to help. Code, benchmarks, a regenerable synthetic corpus, and all workbench traces are released.

End-to-end weather forecasting systems produce skillful global gridded and station forecasts directly from raw Earth observations, replacing the numerical weather prediction pipeline, including data assimilation, at a fraction of its cost. These systems are deterministic and issue no uncertainty. Here we render the Aardvark Weather model probabilistic by attaching one stochastic mechanism to each component: learned, input-dependent noise at the observation encoder, capturing aleatoric uncertainty inherited from the observing system, and Monte Carlo dropout in the processor, capturing epistemic uncertainty in the learned dynamics. The resulting nested ensemble attributes forecast spread to the two sources through a law-of-total-variance decomposition, cross-checked by withholding observation streams. Probabilistic finetuning significantly improves the mean forecast, by 4.2% on average across variables and lead times. The ensemble is calibrated against ERA5 through the medium range (spread-skill ratio 0.98), keeps station RMSE within 2.4% of the deterministic model while beating it in CRPS at every lead time, and trails the operational ECMWF ensemble. The encoder branch behaves as observation-driven uncertainty. Component-attributed uncertainty makes end-to-end forecasts more transparent, a step toward observation-driven digital twins of the atmosphere.

Speech enhancement is often used as a front-end for robust ASR, yet recurrent temporal and cross-band modules introduce sequential dependencies that reduce parallel efficiency. In this paper, we present a sequence-parallel band-split enhancement front-end built on a Parallel Time-Band Mixer (PTBM) block that eliminates within-block recurrent unrolling. PTBM integrates intra-band temporal mixing and per-frame cross-band attention within a unified parallel architecture, enabling efficient contextual modeling across both time and frequency dimensions. The system retains the mask-plus-residual reconstruction interface and introduces learned Observation-Adding (LOA) to suppress ASR-sensitive artifacts without development-set tuning. Experiments on DNS Challenge and CHiME-4 with frozen Whisper back-ends show that the proposed front-end consistently reduces word error rate relative to recurrent band-split baselines while requiring only 0.96 M parameters and 0.58 GMAC/s for the front-end network.

World Models (WM) are increasingly seen as a foundation for intelligent agents that can predict, plan, and act beyond their training distribution. In this paper, we study WMs from a causal perspective across multiple levels of abstraction, ranging from perceptual observations to building a conceptual representation of the structure governing the environment dynamics. We argue that useful WMs must go beyond generative capabilities alone: they should also capture entity properties, entity-to-entity interactions, and entity-to-environment interactions that determine and explain the dynamics of a system. We provide a formal definition of Causal WMs (CWMs) grounded in the tasks they are intended to support, connecting world modelling with existing work in causal representation learning, object-centric learning, causal discovery, structural causal models, and model-based decision-making. Finally, we relate CWMs to the literature on identifiability, clarifying when the components of a WM can be recovered from data and up to which equivalence. With this, we ground WMs in representations and structures that support causal reasoning and informed decision-making.

Global weather reanalyses and forecasts resolve the evolving atmospheric state on coarse grids, but site-specific applications require predictions at arbitrary locations where near-surface conditions also depend on unresolved terrain and land-surface properties. Existing probabilistic downscalers address this gap using hand-crafted topographic and surface descriptors. We ask instead whether Earth observation foundation models can provide transferable subgrid surface representations for probabilistic weather downscaling. We augment a convolutional conditional neural process (ConvCNP) that downscales coarse ERA5 reanalysis fields at ~25 km resolution with a learned local surface descriptor, obtained by compressing a patch of TESSERA embeddings at 10 m resolution. Although these embeddings summarize annual surface conditions, they improve downscaling by encoding persistent surface properties that capture a location's departure from the coarse-grid atmospheric state. Across five climatically diverse regions, the embedding improves point and probabilistic skill at stations held out in both space and time, overall improving CRPS skill by 11.5% for 2 m temperature and 6.2% for 10 m wind speed relative to a topography-only ConvCNP baseline. A hand-crafted descriptor incorporating richer surface information than topography alone captures comparable persistent subgrid signal but yields far smaller predictive gains than the learned embedding representation. These improvements persist when forecasts from the Aurora AI model replace ERA5 reanalysis fields and when predicting at newly deployed weather station networks. To our knowledge, this is the first evidence that long-timescale Earth observation embeddings can support short-timescale weather downscaling where subgrid departures are systematically structured by persistent surface properties.

Multimodal satellite imagery provides complementary information for Earth Observation, but accurately combining heterogeneous sensors remains challenging in dynamic environments. Fast-changing regions, such as the Antarctic marginal ice zone, cannot fully exploit multimodal information from different satellite sensors because surface features move between image acquisitions. This spatial and temporal mismatch challenges effective perceptual grounding, violating the assumption of pixel-level correspondence that underpins most multimodal reasoning and downstream classification pipelines. Antarctic sea ice provides a challenging benchmark due to the rapid, heterogeneous drift of individual ice floes and the differing responses of sea ice to radar, visible and thermal sensing modalities. Accurate, dense supervision of sea ice remains scarce because generating pixel-wise labels requires time-consuming expert interpretation of noisy data, leading to historical reliance on coarse-resolution maritime ice charts for model training. This paper presents a novel architecture based on mutual information warping to align multi-satellite (Sentinel-1 and MODIS platforms) multimodal (visible, thermal, radar) satellite scenes. To demonstrate the approach, we introduce a sparse expert-labeled dataset of 2,088 pixel-wise annotations (7,046 expert point classifications) located at the ice-water margin interface across 43 scenes. Our results demonstrate that spatially grounding and aligning modalities prior to segmentation improves classification accuracy, and enables accurate, dense sea ice segmentation from sparse point-wise supervision.

Bimanual dexterous grasping of large objects is a critical challenge in robotic manipulation. However, most existing studies focus on sequential manipulation rather than cooperative grasping, and methods addressing such bimanual tasks have largely been limited to simulation. These limitations stem from the difficulty of acquiring full 3D object models and generating physically plausible grasping actions. To fill this gap, we propose a real-world bimanual grasping framework that includes: a multimodal dataset capturing joint angles, visual observations and force signals; a Denoising Diffusion Probabilistic Model (DDPM)-based module that generates joint-level grasp configurations from segmented point clouds; and an execution strategy that integrates motion planning with online grasp refinement to ensure physical stability and feasibility. Our approach enables the synthesis of executable bimanual grasps from single-view inputs, reducing dependence on complete 3D object models and ensuring stable real-world performance. Experiments on a dual-arm robot demonstrate high success rates across unseen objects with varying geometries and poses, and ablation studies confirm the contributions of key components of our system.

Medical world models commonly learn from multimodal recordings synchronized onto a fixed-rate grid. This preprocessing resamples each native stream onto a shared time axis. Each stream has an observation clock that governs when observations are emitted or updated. When this clock depends on the latent or acquisition state, it is endogenous. In such settings, synchronization may not be neutral and can erase task-relevant evidence before the model sees the data. We introduce a four-regime taxonomy that characterizes where the evidence needed to distinguish a target event or state survives. The relevant witness may remain in the sampled values, in grid-cell update patterns, in native timing, or only in an external acquisition channel. EndoClock operationalizes this taxonomy as a conservative pretraining audit. It reports the lowest witness-bearing representation supported by the available evidence, or unresolved when no regime can be established. We illustrate this failure in echocardiography, where B-mode video write-outs cease during pulsed-wave Doppler acquisition while the corresponding measurement events remain recorded only in an external acquisition log. This work is a preliminary failure alert and executable audit. Its practical message is to preserve the native observation process long enough to determine whether synchronization has erased information required by the intended task.

Modern Earth observation (EO) satellites carry increasingly advanced sensors that produce vast volumes of high-resolution, multispectral data, yet downlink capacity remains a critical bottleneck -- often causing significant latency or the loss of valuable observations within limited contact windows. We propose a "Summarize First, Download Later" paradigm that exploits recent advances in onboard edge computing and Vision-Language Models (VLMs). Rather than indiscriminately downlinking raw imagery, the system follows a three-phase interaction protocol: the satellite first transmits concise natural language summaries generated by a quantized onboard VLM; ground operators then issue targeted Visual Question Answering (VQA) queries to verify scene relevance (e.g., wildfires or maritime anomalies); and full-resolution images are downloaded only when critical information is confirmed. This transforms the downlink from passive bulk transfer into an active, semantics-aware dialogue. We implement and evaluate the system on a resource-constrained NVIDIA Jetson platform, and experiments on diverse remote sensing scenes show that the proposed strategy substantially reduces bandwidth consumption while accelerating time-to-insight for time-sensitive missions.

Tool-using agents do not merely consume observations: their actions determine what arrives next. In agentic text-to-SQL, a broad query can spend context and database work before useful evidence appears, while post-hoc compression cannot recover omitted rows or expended work. We present BAP-SQL, which treats observation formation as a budget-control stage: it estimates query risk, rewrites SQL when useful, and delegates hard limits to an independent runtime shield. Across general 4B, specialized FINER-SQL 4B, and 7B backbones, BAP-SQL improves tight-budget success. On the primary BIRD-derived setting, it gains 3.4/3.6 percentage points over matched SFT while using 4.5/5.0% fewer tokens. Matched retraining and task-level transfer associate the gain with policy-visible planning and budget-sensitive rescue. The benefit attenuates as model capability and budget increase, reverses at the loosest setting, and does not reduce database work.

A calibrated stochastic world model can reveal how uncertain a future is without revealing why it branches. The same conditional future law can arise because an observation aliases physical states or because dynamics remain random after the declared full state is fixed. We prove that ordinary transitions cannot identify these two sources, even for a perfect probabilistic predictor. ClosurePairs makes them identifiable by crossing compatible microstates with repeated exogenous disturbances and estimating state, noise, and state-noise interaction variance. The central consequence is operational: under finite hierarchical sampling, forecast difficulty governs the useful compute scale, while the alias/process composition provides complementary information about its direction-resolving the current state or sampling future randomness. ClosurePairs recovers source attribution at unchanged likelihood, reduces equal-budget decomposition error in a nonlinear interaction benchmark, and supports observation-only routing. On exact-marginal MetaWorld twins, an output-only allocator is at chance while a Closure-supervised probe on frozen JEPA-WM features routes 89.8-100%. In an independent ManiSkill PushCube confirmation, a stochastic RSSM's outputs and latents remain at chance, whereas an RGB-only Closure probe routes 100% under both ID and geometry/camera OOD over five seeds, matching direct allocation rather than exceeding it. Across five unseen allocation menus, the same Closure probe routes 92.5%/90.4% ID/OOD with no new oracle labels, versus 37.9%/32.9% for a frozen direct allocator. ClosurePairs is therefore an identifiable, reusable mechanism target that cannot be recovered from forecast quality alone.

4D Gaussian Splatting (4DGS) can render dynamic scenes photorealistically. However, with limited viewpoint coverage, some spatiotemporal regions remain sparsely observed, leading to artifacts, particularly in scenes with large motion. Existing approaches leveraging generative models rely on heuristic virtual-viewpoint selection before refining rendered views. As a result, they cannot actively explore such sparsely observed regions. To address this issue, we propose a pipeline that actively selects spatiotemporal virtual viewpoints to improve 4DGS reconstruction. Our method selects virtual viewpoints for generative enhancement based on the rendering sensitivity and motion-aware observation density of 4D Gaussians, prioritizing views that alleviate observation sparsity. In the refined images, we filter out regions that conflict with captured observations or are likely to contain generative artifacts and then fine-tune 4DGS using only the reliable regions. We evaluate our method on multi-view video benchmarks using new train/test splits designed to induce observation gaps. Results show consistent improvements over prior viewpoint selection strategies and fine-tuning methods in both qualitative and quantitative evaluations, while reducing artifacts.

Building interactive digital twins requires recovering both 3D geometry and the kinematic structures that govern how objects articulate. Yet existing methods for articulated object reconstruction require explicitly observable motion from multiple articulation states. We introduce a rest-state formulation that reconstructs articulated objects from a single closed configuration, an inherently ill-posed setting where geometry, semantics, and motion priors compensate for the absence of motion cues. Our framework adopts an explicit mesh as an intermediate representation for cross-model verification and fusion, reconciling noisy outputs from vision-language and segmentation models into spatially consistent part structures. To estimate joint parameters without observed motion, we use a video diffusion model to synthesize articulation hypotheses and validate them through geometric consistency. Our approach achieves accurate part decomposition and physically plausible articulation, performing competitively with motion-observing reconstruction-based, generation-based, and modular pretrained-model baselines.

While MoE augments VLA via expert specialization, router suffers from ineffective expert routing owing to the kinematic heterogeneity of actions across manipulation tasks and, even worse, the unavailability of the kinematic signals at inference time. In this work, we first observe that most semantically distinct manipulation tasks reduce to multiple kinematic archetypes. Motivated by this finding, we propose Kinematics-supervised explicit routing (KinRT), a new paradigm that shifts from implicit, observation-driven expert routing to explicit, kinematics-guided expert dispatching. Specifically, we perform kinematic clustering on action trajectories into multiple kinematically coherent groups, whose IDs serve as ground truth to supervise the training of the router; at inference time, the router dispatches experts only using visual-language observations, without any reliance on action kinematics. KinRT actually introduces an asymmetric bridging mechanism that distills the task kinematics from the action space in training into the observation space at inference. In addition, to assess KinRT's cross-platform generalization, we build an economical, Do-It-Yourself robot (DIYRobot) platform from scratch using 3D-print technology ($<$ 2,000USD). Extensive experiments demonstrate KinRT's superiority over both dense and MoE-featured VLAs by more than 23.26% on RoboTwin benchmark and 20.27% on our introduced DIYRobot platform. Our code and DIYRobot platform will be open-sourced.

Detecting when the statistical behavior of an engineered system changes, and identifying which component is responsible, are core problems in the monitoring of telecommunication networks, robotic platforms, security infrastructure, and multi-agent systems. In safety- and mission-critical deployments, such decisions must be accompanied by statistical reliability guarantees rather than by point estimates alone. Conformal changepoint localization (CONCH) and conformal root cause analysis (CROC) meet this need by returning confidence sets that contain the true changepoint, or the true root-cause stream, with a user-specified probability, without parametric assumptions on the data-generating process. In practice, however, observations are frequently corrupted, e.g., by outliers, sensor faults, or adversarial perturbations. While the finite-sample coverage of these procedures is preserved under contamination, the resulting confidence sets can become uninformatively large. Adopting a Huber-type contamination model, this paper proposes weighted CONCH (W-CONCH) and weighted CROC (W-CROC), which downweight observations that are likely to be corrupted with the goal of reducing confidence set size when data may be corrupted. The weighting mechanism, derived from a formal bound on the unknown corrupted data densities, leverages pre-existing second-order classifier-based uncertainty signals, such as those produced by evidential deep learning or Bayesian learning. W-CONCH and W-CROC are further generalized by introducing a meta-learning procedure for the weights that optimizes a differentiable surrogate of the confidence set size. Experiments on image-based and real-world changepoint and root-cause benchmarks show that uncertainty-based weighting substantially reduces confidence set size while maintaining the target coverage.

AI agents are increasingly adept at tackling complex, long-running tasks. With the rapid surge of autonomous capabilities, human oversight is systematically lagging behind due to limited human-centered interfacing. Aiming to address this, we introduce AgentGUI, a user-friendly, locally hosted GUI for seamlessly observing and steering AI agents amid multiple concurrent, long-running sessions. AgentGUI features 1) rich agent trajectory visualizations, 2) effective manual and automated steering, and 3) integration with and coordination between open-source and frontier agent frameworks. A controlled user study demonstrates statistically significant reduction in the time it takes to identify key elements from agent traces (38% faster, p = 0.023). In a preliminary experiment, AgentGUI's automated drift prevention feature raises the task completion rate of small local agents by as high as 34pp across a 0.8B--9B model ladder (N=50 runs per model). AgentGUI is publicly available through its project website (https://agent-gui-project.github.io) and open-source repository (https://github.com/eth-medical-ai-lab/agent-gui), along with a demo video (https://youtube.com/watch?v=GSDyxN1gTF0).

Brains and large language models (LLMs) are fundamentally different memory systems, but they can be compared through shared functional questions: where memory-related information is represented, how partial cues recover broader associations, how new information is written or updated, and how memory-related states can be perturbed. In biological systems, these questions span synapses, neuronal ensembles, hippocampal-cortical interactions, and plasticity; in LLMs, they span weights, activations, context windows, retrieval systems, and external stores. The comparison is therefore functional and experimental rather than anatomical. Human studies reveal sparse concept responses, temporal binding, rapid association formation, episode-specific coding, and recall-related reactivation, but selective intervention remains limited. Rodent studies provide more selective causal access to learning-related ensembles, whereas human and macaque interventions usually affect broader circuits. LLMs lack lived episodic memory, yet they permit unusually direct and repeatable manipulation of internal states and stored information. We argue that this asymmetry creates a new opportunity. LLMs are not ahead in memory itself, but in experimental access. Their tools may help turn broad questions about retrieval, updating, persistence, reversibility, and unintended effects into sharper biological hypotheses. The productive bridge is to transfer experimental logic, not anatomical parts.

We introduce time-reversed imaging, a new paradigm that infers what just happened in a scene from fading multimodal traces. Instead of extrapolating or interpolating video frames, our goal is to infer past human-environment interactions from residual physical imprints observable in thermal, ultraviolet, and visible spectra. To study this problem, we present TRACE-HEI, the first proof-of-concept dataset for time-reversed imaging, containing synchronized tri-modal video sequences of actions such as sitting, touching, moving objects, and liquid spills, captured across diverse materials and recorded up to three minutes after contact. To establish the benchmark, we propose a multimodal inference approach that extracts structured textual descriptions of detected traces and uses them to constrain a vision-language-guided diffusion model for reconstructing plausible past frames. Experiments show that inferring recent events from fading traces is challenging but feasible when complementary modalities reduce solution ambiguity. This work defines the first computational and experimental foundation for time-reversed imaging, bridging vision, physics, and generative reasoning, and opening new directions for scene understanding beyond instantaneous observation.

Affective Image Content Analysis (AICA) aims to recognize and understand emotions elicited by visual content, representing an indispensable step toward Artificial General Intelligence (AGI). However, despite the rapid progress of Multimodal Large Language Models (MLLMs), systematic evaluation of their visual emotional intelligence remains largely absent from recent model releases. We attribute this gap to a structural mismatch between conventional AICA paradigms and the open-ended, instruction-driven nature of MLLMs, where further analysis reveals four major limitations: omission of plausible responses, limited emotion taxonomies, neglect of contextual factors, and labor-intensive annotation. To overcome these barriers, we introduce Emotion Statement Judgement (ESJ), a statement-verification formulation that preserves the expressiveness of the input space while constraining outputs to discriminative judgements. We further develop INSETS, a labor-efficient pipeline that instantiates ESJ at scale by constructing INSETS-462k and supporting MVEI, a rigorously refined benchmark spanning sentiment polarity, emotion interpretation, scene context, and perception subjectivity. Beyond evaluation, we build EmObserver, an emotion-oriented MLLM optimized on ESJ through an elaborate multi-stage recipe. Extensive evaluation of broad-spectrum MLLMs on MVEI reveals fine-grained insights into current artificial visual emotional intelligence, while experiments on multiple AICA benchmarks demonstrate the accuracy, generalization, and reasoning faithfulness of EmObserver. Collectively, these results establish ESJ as a practical formulation, MVEI as a comprehensive benchmark, and EmObserver as an advanced baseline for advancing MLLM-oriented visual emotional intelligence. Code will be released at: https://github.com/wdqqdw/EmObserver.

Data-driven methods have revolutionized ocean modeling, yet current approaches rely heavily on complete reanalysis datasets, imposing computational constraints and limiting model performance to that of the training data. Here, we present a generative state-space model and an optimization framework that enable learning directly from sparse and noisy observations. The model is essentially a hidden Markov model with a continuous state space, where oceanic physical quantities are treated as hidden states and measurements as observations, enabling a unified representation of ocean fields and observational data. Both the initial-state and state-transition modules are implemented as neural networks to capture the complexity and temporal evolution of ocean states, while the emission module is formulated as a masked Gaussian distribution. To train the model from sparse observations, we derive an optimization framework based on the expectation-maximization (EM) algorithm. The framework alternately reconstructs high-fidelity ocean fields via Langevin dynamics and optimizes deep neural networks to capture temporal evolution. Theoretical analysis shows that the framework maximizes the likelihood of observations under the generative model. For efficiency, we assume that ocean-state evolution follows a stationary, ergodic, and Markovian stochastic process and adopt only length-two state sequences during optimization. Experiments on CMIP6 simulation data and FY-3D satellite data demonstrate high-fidelity reconstruction and accurate prediction, showing that sparse observations can directly improve the model's representation of ocean-state dynamics. This work offers a scalable pathway for next-generation Earth system models to learn directly from sparse, incomplete real-world observations.

Managing massive crowds during infrequent special events requires reliable real-time pedestrian-flow forecasting to ensure public safety and operational efficiency. However, supervised forecasting methods face limitations in these contexts due to scarce historical data, heterogeneous data distributions, and short in-event observation windows. To effectively support operational decision-making, forecasts should provide not only accurate point estimates but also informative predictive uncertainty. Probabilistic uncertainty quantification plays a critical role in this aspect, particularly capturing sudden volatility and tail risks. This paper investigates pretrained time series foundation models as a lightweight approach for zero-shot probabilistic forecasting without extensive local retraining. Using decision-oriented metrics tailored to short events, we conduct a comprehensive assessment of two time series foundation models on crowd forecasting, with the SAIL2025 event as a use case. We then distill practical insights for crowd managers, specifying when zero-shot forecasts remain operationally reliable.

Human vision is a closed loop: gaze is continuously redirected by intermediate hypotheses rather than a single snapshot. Decades of psychophysics and cognitive science have argued that this active observation is essential for a wide range of tasks. Whether today's multimodal large language models (MLLMs) exercise active observation is an empirical question that current vision-language benchmarks do not answer. We introduce ActiveVision, a benchmark that makes active observation measurable for MLLMs, comprising 17 tasks across 3 categories. Tasks are designed to force repeated visual perception rather than a single static description. Frontier MLLMs collapse on ActiveVision: the highest-scoring model we evaluate, GPT-5.5 at the highest exposed reasoning-effort tier, solves only 10.6% of items and scores zero on 11 of the 17 tasks, and even Claude Fable 5, despite topping most reasoning and coding leaderboards, solves just 3.5%, far behind three human participants who average 96.1%. Furthermore, much of the gap persists even when models write and run their own vision code: such code is unreliable on realistic imagery, and catching its failures itself requires the active perception the models lack. Together, these results indicate that current MLLMs lack robust active visual observation, motivating architectures and training objectives that close the perception-reasoning loop.

Inverse Reinforcement Learning (IRL) algorithms are powerful tools for learning from and generalizing expert demonstrations, but they often rely on unconstrained exploration, rendering them unsafe for real-world deployment. Meanwhile, Control Barrier Functions (CBFs) can guarantee the safety of control systems, but the analytical design of CBFs can be time-consuming and esoteric. In this work, we address these limitations jointly by constraining reward function candidacy during IRL to the space of CBFs, yielding a formulation that exhibits safe online control with continuous experiential improvement. Crucially, this framework enables the data-driven recovery of barrier functions directly from unlabeled expert observations. We demonstrate that the recovered barrier function is robust to unsafe states entirely absent from the expert data. Furthermore, we benchmark our method against standard IRL baselines in a simulated navigation environment, demonstrating improved safety performance. Finally, we investigate the trade-offs of planning-based versus policy-based IRL methods across both simulation and a real world obstacle avoidance task.

The bottleneck of Earth Observation processing chains is not the arrival of new imagery but whether the surface is actually visible when the image arrives. We study this as an observability forecasting problem on EarthNet2021. Given recent multispectral imagery and exogenous weather drivers, the goal is to predict whether the next acquisition will be usable and, if not, when a usable view is likely to return. To do this, we adapt LeWorldModel, a joint-embedding predictive architecture world model, to cloud-aware Earth Observation sequences. The final pipeline converts raw minicubes into episodic HDF5 sequences with five image channels (blue, green, red, near-infrared, cloud mask) and eight meteorological and calendar covariates. The resulting model has 18.0M trainable parameters and is trained from scratch on 23,904 training episodes. The trained leWorldModel is evaluated under a locked protocol: linear probes are fit on train only, calibration choices are set on an internal validation split, and the fitted heads are then frozen for valsplit, IID, OOD, and extreme evaluation. On the full frozen-bundle observability benchmark, LeWorldModel consistently outperforms persistence. For next-step usability, balanced accuracy ranges from 0.769 to 0.887, compared with 0.493 to 0.556 for persistence. For exact first-usable-horizon prediction, accuracy ranges from 0.602 to 0.806, compared with 0.120 to 0.369 for persistence. Against a frozen LightGBM baseline fit on the same training windows, LeWorldModel is better on continuous clear/cloud regression and on exact recovery timing on valsplit, IID, and extreme, while LightGBM is stronger on the simpler binary any-usable-within-six task and is more robust on OOD. In separate sampled diagnostic analyses, LeWM also produces strong ranking-based anomaly signals under synthetic temporal inconsistencies.

Recent advances in foundation models have transformed AI for Science, enabling remarkably accurate predictive performance across domains ranging from protein folding to weather forecasting. Yet prediction alone does not constitute scientific discovery. Scientific understanding depends on uncovering the reusable explanatory mechanisms that generate observations, whereas contemporary machine learning remains fundamentally organised around predictive mappings rather than explanatory structure. In this paper, we argue that scientific discovery is fundamentally a problem of knowledge organisation. To this end, we introduce Mechanistic World Models, a new design paradigm that places reusable mechanisms at the centre of representation, computation and learning. Drawing on insights from the philosophy of science, we derive the computational capabilities required for discovery, identify the design principles and inductive pressures that encourage explanatory knowledge to emerge, and formalise the anatomy of a mechanism-centric world model. Finally, we show how diverse research directions including mechanistic interpretability, causal representation learning, equation discovery and modular architectures capture complementary ingredients of this paradigm while lacking a unified framework. We propose Mechanistic World Models as a conceptual foundation and computational blueprint for moving AI beyond predictive forecasting towards autonomous scientific discovery.

A specialist tolerates blind spots that a generalist does not. Usually this is treated as a cost to be minimized. We treat it as a design variable: a deficiency can be kept because it pays and removed on demand in the rare situation where it would be fatal, by routing to a compensation channel. We give three results. First, an advantage condition under which keeping the deficiency is a computable economic position; structurally it is the Ehrlich-Becker market-vs-self-insurance margin applied to a competence gap, with the detector as a Townsend costly-state-verification technology. Second, a two-sided characterization of removability. A coupling lemma shows that when the deficiency is a coarsening of perception, no switch can separate benefit from harm, yielding a converse (a confounded detector earns zero premium, and any within-defect policy insisting on positive premium is driven, under multiplicative dynamics, to negative long-run growth) and an achievability result (a detector outside the deficiency earns a positive premium). Together, over structured uncertainty classes with severity capped or miss rate O(1/L): a defect is profitably removable iff the detector-relevant distinction survives the restriction and the advantage condition holds; the premium is the support function of the class's ROC set at an economic price vector. Third, observation defects and capacity defects differ exactly on whether access to the deployment distribution rescues them; the gap decomposes as cross-leak plus a closure deficit, and per-task randomization buys back the latter, never the former. The detector can be learned from declared fatal categories at a training bill linear in loss severity (up to a log factor). The results synthesize Chow's reject option, Kelly growth under ruin, and selective prediction.

Foundation models (FMs) have transformed machine learning from isolated task-specific model development toward general-purpose models pretrained on broad data and adapted to multiple downstream tasks. Earth observation (EO) is an important domain for this paradigm because satellite and airborne archives are large, high-revisit, and increasingly multimodal, while reliable field labels are often sparse. Remote sensing foundation models (RSFMs) cannot be transferred reliably/optimally without domain-specific adaptation. This is because EO data are governed by measurement physics and operational decision constraints. This chapter reviews the design principles arising from these domain-specific constraints. It first defines the FMs paradigm in remote sensing (RS), then synthesizes the current model landscape, pretraining objectives, architecture designs, downstream adaptation and trustworthiness requirements. The chapter also incorporates recent benchmark evidence showing that no single geospatial foundation model is universally best and that inconsistent evaluation remains a major issue to fair comparison and reliable deployment. In addition, two brief environmental monitoring case studies; physics-informed spectral targeted masking for harmful algal bloom prediction and reinforcement learning for adaptive environmental monitoring station selection to illustrate the FMs domain-guided principles in practice. This chapter posits that next-generation RSFMs should be evaluated not only by benchmark accuracy, but also by modality-aware transfer and physically plausible representations for trustworthy EO decisions.

The Internal Waves Service screens the Sentinel-1 Wave-mode archive for internal solitary waves, routing detections to experts whose adjudication time is the resource the effort exists to conserve. Because attention is the cost of error, precision leads. Its classifier was trained and reported at a one-to-one class balance, fixed before the operational rate could be known. That rate has since emerged at roughly one scene in twenty, and a balanced-test score badly overstates the precision a validator meets. A model that scores 0.794 balanced-test precision scores 0.192 in real operation: the gap is a systematic artefact of reporting at the wrong prior, invisible to the metric most work quotes. We show the mismatch to be an evaluation problem in the costume of a training one at a fixed recall, prior correction and calibration cannot move precision, and answer it with a prior-matched reporting method based on three numbers: balanced-test, operational-prior, and real post-deployment, whose contrast is the honest measure. A precision-first, leakage-controlled development cycle then improves the classifier lever by lever, each promoted only against a pre-registered margin; negative variety and the aggregation head lifting, capacity paying once then stopping, calibration inert, so the honest negatives are as much a result as the gains. Holding recall at a floor of 0.80 and certifying against a sealed, single-read lockbox, the promoted model reports 0.927 precision at the operational prior; an out-of-time check confirms discrimination transfers to unseen periods while a fixed operating point does not. Prior-matched reporting, begin balanced, then move to the prior as the stream reveals it, transfers to any operational Earth-observation service bootstrapping a rare-event detector under a prior it has yet to discover.

Reliable calibration of multi-fisheye camera systems remains challenging as rig size, camera arrangement diversity, and field of view increase. Existing pipelines can jointly optimize intrinsics, extrinsics, and target poses, but their success still depends heavily on empirical capture rules and the quality of the observations supplied to the solver. This paper studies this dependency through a failure-oriented analysis. We reveal that calibration failures are not sufficiently explained by detector recall loss or global image-plane distribution imbalance. Instead, the dominant failure factor lies in intrinsic initialization: observations with limited radial span couple focal scale with fisheye projection-shape parameters, producing ill-conditioned updates. Guided by this insight, we propose CO-Calib, a plug-in calibration-data construction framework that combines a robust learning-based target detector with an error-analysis-guided frame selector. CO-Calib constructs initialization-friendly anchors, co-visible multi-camera constraints, and coverage-completion frames without changing the existing calibration workflow or optimization backend. Extensive experiments on synthetic and real multi-fisheye systems demonstrate that CO-Calib improves the overall success rate from 68.1% to 99.3%, increases extrinsic accuracy, and augments real-world calibration stability. The source code will be made publicly available at https://github.com/HKUST-Aerial-Robotics/CO-Calib.

Navigational bronchoscopy is critical for pulmonary interventions, yet current platforms depend heavily on pre-operative CT or external sensors, limiting their use in critical care and resource-constrained settings. Vision-only navigation offers a scalable alternative, but conventional visual odometry (VO) struggles with texture-poor airway images, specularities, and the vanishing-point singularities of tubular anatomy, leading to frequent tracking failures and drift. We present a geometry-aware VO framework that explicitly leverages vanishing-point cues from airway lumens. Detected lumens are back-projected to 3D rays, whose weighted fusion yields a stable forward heading even when parallax cues are absent. This heading, together with looming-based velocity estimates, is fused with noisy VO outputs using a bespoke high-gain observer that enforces airway-following priors and rejects drift. We validate the method on ex-vivo mechanically ventilated human lungs with electromagnetic tracking ground truth. Compared to state-of-the-art pipelines (ORB-SLAM2, LoFTR-VO, DPVO), our approach reduces absolute trajectory error by more than 50% and achieves the lowest relative pose error across all test sequences.

Sampling action chunks via generative models has become a widely adopted methodology for robotic learning from demonstration. However, existing methods often struggle to balance responsiveness and computational cost because they execute each action chunk for a fixed execution horizon. In this paper, we adaptively adjust the execution horizon of sampled action chunks, balancing responsiveness and computational efficiency. We introduce Spatial Attention -- defined as the expected squared norm of the gradient of the action log-likelihood with respect to the observation -- which indicates the sensitivity of the policy's action distribution to variations in the observation. We show that, under a fixed budget of chunk samplings, the execution horizon that minimizes the cumulative likelihood drop induced by disturbances decreases as Spatial Attention increases. By forecasting future Spatial Attention values alongside the action chunk, our framework dynamically assigns shorter execution horizons to phases with high Spatial Attention, and longer horizons to phases with low Spatial Attention. Experiments on standard and perturbed tasks, in both simulation and on a real robot, show that our method significantly improves success rates over fixed-horizon baselines while maintaining the average execution horizon.

When evaluator-derived step rewards are pooled or compared across scoring channels, their sign is treated as transportable. Yet the same frozen transition can change sign with the scoring channel. Process rewards vary agent states, while evaluator audits vary scoring configurations; neither first difference isolates their interaction. We define Agent Step Value (ASV) as channel-indexed target-margin gain and identify the missing state-by-channel interaction by replaying complete faces. Of 1,100 PubMed open question-answering transitions, 1,004 were complete across four cyclic layouts. Their mean update is positive under direct scoring (+0.163 [0.102, 0.218]) and negative with an externally generated view (-0.160 [-0.244, -0.079]), giving an interaction of -0.323 [-0.418, -0.232]. Across paired transitions, 507/1,004 cross zero in one direction or the other. Matched templates isolate a generated-minus-quote interaction of -1.121 [-1.534, -0.703]. Its direction remains negative after the readout and evaluator-stack changes, and quote yields a higher paired area under the receiver operating characteristic curve (AUC) for stored success at all three bridge vertices. Because the two views preserve the same task information conditional on the retained state, a representation-invariance null predicts equal responses and success rankings across them. ASV provides a transport audit for evaluator-derived step measurements; causal actor credit lies outside its estimand.

High-throughput plant phenotyping generates valuable data that often remains trapped in unstructured text and isolated RGB images. To bridge this semantic gap, we propose a framework for constructing a multimodal granular Knowledge Graph (KG) to monitor genotype-phenotype interactions across time and experiments. In this work, we focus on wheat Triticum aestivum as a representative target crop to validate our methodology across complex canopy environments. Our pipeline first distills noisy field notes to extract entities and relations, dynamically constructing the KG by converting unique instances into hierarchical class entities via RDF-typing. These graph nodes are then aligned with standardized ontologies (PO, RO, WTO) using PlantDeBERTa. To visually ground the constructed graph, a Vision-Language Model paired with a wheat-segmentation ViT generates attention-based softmaps, linking specific KG entities directly to image pixels. We introduce a central observation node Plant_Obs_Id to connect these multimodal subgraphs temporally. Evaluated on 500 curated WisWheat samples using Pointing Game accuracy, Visual Word Sense Disambiguation (VWSD), and rank-based metrics, our neuro-symbolic approach successfully maps complex field observations to a structured graph. This enables automated field note auditing, temporal stress monitoring, and precise spatial trait localization for wheat breeders.

Post-hoc analysis of trained neural network weights often seeks to recover geometric structure directly from the parameters. We show that, for positional-encoding-equipped neural fields, the symmetry visible from weights is not the true symmetry group itself, but an observable symmetry set determined by the trained parameters, the positional encoding (PE), and readout observable. We formulate this dependence through an exact observability hierarchy, $G_{\mathrm{obs}}^{\mathrm{exact}} \subseteq G_{\mathrm{lift}}^{\mathrm{exact}}(φ) \cap G_{\mathrm{true}}$, where $G_{\mathrm{lift}}^{\mathrm{exact}}(φ)$ is the set of input transformations that the PE can exactly lift to the feature space. The hierarchy implies that even when a target function has a geometric symmetry, that symmetry may be structurally invisible to weight-level observables if the PE does not represent the corresponding transformation. We test this prediction using MLPs trained on two-dimensional signed distance functions with multiple shape symmetry groups, positional encodings, and Gram-based observables. The results show a consistent PE-dependent pattern: DyadicAxisPE supports $D_4$-sensitive readout but structurally suppresses $D_3$ rotations, TriAxisPE yields lower $D_3$ / $D_6$ readout scores under the tested Gram observables by replacing coordinate axes with three 120-degree-separated axes, and random Fourier features mainly exhibit a $π$-rotation response under these readouts. These findings show that PE design affects not only approximation behavior but also which structures are accessible to post-hoc weight-level readouts. This provides a basis for a principled observable-dependent symmetry readout.

NASA and its data centers hold thousands of geoscience datasets and tools like Worldview, Giovanni, the Science Discovery Engine, and Harmony. Finding the right one is hard even for domain experts. We present an agentic search system, deployed as a public service for the geoscience community, that takes a natural-language research query and returns the matching datasets and tools. We demonstrate that, in the era of large language models, the latent value of knowledge graphs (KGs) can be substantially amplified through agentic search. From the NASA Earth Observation Knowledge Graph (NASA EO-KG) we derive NASA-EO-Bench, an open benchmark of 47k query-dataset pairs (21k task-based queries). A neural scorer fine-tuned on NASA-EO-Bench beats cosine and BM25 baselines. Further combining it with BM25 via score fusion raises both Recall@10 (R@10) and MRR by over 5x. On top of this supervised pipeline, we add a zero-shot agentic reranking stage that, without any additional training, lifts MRR by 28% on a stratified N=200 subset, showing that LLM reasoning is complementary to supervised retrieval.

How many days of early behavior suffice for subscription churn prediction? In the public KKBox dataset, the early indicator of churn is typically an indicator of someone's contract status; however, when looking in the heavily churned manual-renewal segment, having access to early behavior creates a substantial increase in prediction for that specific segment (PR +0.10 at 120 days). A nine-window sufficiency curve shows diminishing returns; descriptive inflection detectors place a knee between 30 and 120 days depending on the metric and model class. Stress-testing over three cohort/task designs shows that this curve is singular to the design being tested; for example, in our test with a moving target, the curve inverts and can shift depending on the feature set used. Therefore, any window-sufficiency claim should state its cohort construction, target definition, and feature families. A survival-filtered replication on educational-attrition data (OULAD) confirms the rising-curve shape in one behavior-dominated setting (floor ROC 0.56 vs. 0.88), though signal composition inverts.

LiDAR semantic segmentation often degrades under real-world deployment due to evolving sensing conditions, while collecting new annotations for retraining is impractical. Test-time adaptation (TTA) updates model parameters online using pseudo-label supervision, but directly applying standard TTA strategies to LiDAR data is challenging. Because pseudo-label reliability is spatially heteroscedastic under range-dependent sparsity and occlusion, uniform updates on globally shared parameters can inject unstable gradients and destabilize adaptation. We propose a geometry-constrained test-time prompt tuning framework for LiDAR semantic segmentation. Our method estimates per-location sensing reliability from depth-consistent beam terminations and neighborhood support, and uses it to reweight spatial supervision. Adaptation is confined to lightweight prompt adapters inserted into a frozen backbone, with spatial gating to prevent unreliable regions from perturbing globally shared representations. A temporally smoothed prototype alignment strategy further stabilizes online updates by accumulating reliable semantic evidence over time. Experiments on standard LiDAR benchmarks demonstrate improved adaptation stability and segmentation performance under deployment variations without additional annotations.

We consider sparse multivariate stochastic systems that evolve in continuous time according to a causal mechanism and present methodology to recover the system's time-infinitesimal transition mechanism from mere cross-sectional data. This observational paradigm is motivated by applications such as gene expression analysis, where destructive experimental techniques may only allow recording data once over a cell's lifetime. Precisely, we assume the system follows a time-homogeneous diffusion process that has reached an equilibrium distribution at observation time. Further, we assume the causal mechanism is fully described by the diffusion drift, is acyclic, and its causal structure graph is known. In this setting, we prove that the full causal mechanism, i.e., the drift function, can be non-parametrically identified under a weak non-explosion criterion. We derive a non-parametric kernel estimator for this challenging inverse problem and prove its consistency. Moreover, we propose a cross-validation scheme for hyperparameter tuning, illustrate the behavior of our estimator in simulations, and we discuss connections with irreversible generative diffusion models and low-frequency sampled data.

SWE-agent established the action interface as an underexplored design axis for software-engineering agents; we make the analogous case for the observation interface in computer-use (CU) agents. Current CU agents, closed and open-source alike, tie observation to action--one screenshot every 3-5 s, no audio--leaving them blind and deaf between screenshots to video, animations, transient UI events, meetings, and spoken instructions. We introduce the Agent-Computer Observation Interface (AOI), a model-agnostic perception layer that decouples continuous, adaptive observation from discrete actions through three gated components: inter-step keyframe capture, volume-gated audio transcription, and CU-model-generated visual narration that persists as text. Each produces almost nothing on static, silent content, reducing to the standard loop without degrading it. On DynaCU-Bench (100 dynamic browser tasks plus a 50-task static control), CU models from 7B to frontier scale gain +17 to +48 pp over their screenshot baselines with zero retraining, turning tasks that are near-impossible from periodic screenshots into largely solved ones. The gap is starkest on audio: on a spoken-content subset AOI agents solve every task, whereas streaming voice models hear accurately but cannot act on what they hear without the scaffold. The decomposition is as informative as the headline gain: keyframe selection turns out not to matter--the value comes from narrating captured frames into persistent text--and the interface is not a fixed bundle, since on a newer model (Gemini 3 Flash) the keyframe stream actively regresses through image-token dilution, so its components must be selected per model rather than shipped as one configuration.

Embodied reasoning and exploration are increasingly considered crucial abilities for robots operating in complex and unfamiliar environments. To accomplish tasks in such settings, an agent must identify and acquire the information necessary for the task through exploration. We propose ObsGraph, an observation-centric hierarchical scene graph that unifies scene representation, retrieval, and exploration. It retains visual evidence and organizes it into room-view-object layers: rooms provide coarse semantic anchors, views preserve contextual object covisibility, and objects store fine-grained details. On top of this representation, we perform coarse-to-fine hierarchical retrieval under a bounded budget, and crucially use retrieval outcomes to structure the exploration candidate space--activating room-level exploration, view refinement, or frontier exploration--thereby tightly coupling representation, retrieval, and adaptive multi-scale exploration. Experiments across embodied reasoning and exploration benchmarks demonstrate improved success and efficiency, highlighting the benefits of structured scene representation and more targeted information gathering driven by identified evidence gaps.

Augmenting model-free reinforcement learning (RL) with representations learned through observation dynamics prediction (observation-predictive RL) can improve sample efficiency and performance, with minor modifications and limited additional computation. However, this approach still struggles in challenging tasks with low-dimensional observations. In this paper, we identify a key factor behind this problem: unbalanced reconstruction losses across observation dimensions, where dimensions with larger value ranges dominate the loss. This encourages the agent to neglect dimensions with relatively small ranges, leading to degraded performance. To address this issue, we propose a novel normalization method tailored to online RL, which normalizes low-dimensional observations and balances the resulting losses and gradients. Beyond balancing reconstruction losses, observation normalization enables dynamics prediction to be performed in a normalized observation space, thereby providing a unified treatment of low- and high-dimensional inputs (e.g., physical states and images). Building on this idea, we further introduce Normalized Observation Space Dynamics-Augmented Q-learning (NASDAQ), a framework for observation-predictive RL applicable across diverse domains. NASDAQ learns state-action representations by coupling value learning with two auxiliary tasks: short-term value prediction and next normalized observation prediction. Extensive experiments demonstrate that NASDAQ achieves competitive or superior performance compared with state-of-the-art model-based and self-predictive RL methods, while requiring significantly less training wall-time.

Deploying multimodal foundation models as closed-loop policies increasingly requires conditioning actions on observations that are no longer visible. However, existing benchmarks either expose the full state, conflate hidden-state reconstruction with other agent skills, or test recall only after an episode has ended. We introduce RNG-Bench (Reconstructive Non-Markov Games), a benchmark suite designed to isolate a base model's ability to reconstruct past observations and act on them during multi-step interaction. RNG-Bench includes two complementary games: Matching Pairs, where card identities briefly revealed at specific locations must later be recalled, and 3D Maze, where egocentric views must be integrated into a spatial map. Both games are evaluated under a unified harness with three controlled difficulty axes: grid size, visual pattern, and observation modality. The benchmark further introduces a head-to-head duel protocol to control for instance-level variance and a Memory Gap metric that disentangles forgetting from poor action selection. The hardest configurations require contexts of roughly 128K tokens and 350 image inputs per episode, and remain far from saturated by frontier MLLMs. Memory Gap analysis shows that most residual errors stem from forgetting earlier observations rather than from suboptimal decision making. Finally, fine-tuning Qwen3.5-9B on optimal-policy rollouts and filtered model demonstrations improves performance on RNG-Bench and transfers to existing benchmarks without degrading general multimodal capability.

Vision foundation tools such as open-vocabulary detectors, segmentation models, and post-processing operators are powerful building blocks for computer vision, but their effectiveness depends heavily on how they are orchestrated: which tools are used, in what order, with what parameters, and under what visual conditions. Existing visual-programming agents typically generate a fixed solution pipeline, making them brittle under dense objects, occlusion, small targets, and domain shift. We introduce VTOS (Vision Tools Orchestration Search), a framework for adaptive visual tool orchestration through joint solution-observer search. VTOS co-searches executable solution programs that compose vision tools such as Grounding DINO, SAM, NMS, and slice-and-detect, together with observer programs that diagnose candidate solutions, identify failure modes, and generate actionable feedback. These observations are accumulated in a shared VisionThoughts knowledge base to guide subsequent search. We evaluate VTOS through two case studies: dense object counting on LVIS-Count and zero-shot plant-disease segmentation on PlantSeg-OOD, which stress different orchestration challenges including threshold calibration, NMS, slicing, mask refinement, and domain generalization. Across both tasks, VTOS outperforms static tool pipelines and agentic visual-programming baselines, specifically in complex settings such as dense, occluded scenes and out-of-distribution segmentation where static pipelines leave measurable headroom, rather than in standard tasks where a single well-calibrated tool already approaches its ceiling.

With sophisticated cyber-attacks becoming increasingly prevalent, modern networks require intelligent autonomous cyber-defense agents trained via Reinforcement Learning (RL). These agents employ neurosymbolic approaches such as behavior trees with learning-enabled components (LECs) to learn, reason, adapt, and implement security rules while maintaining critical operations. However, these autonomous networks are partially observable systems, i.e., the cyber-attacker's (red agent's) actions are not observable, making it difficult for the defender to predict red actions, learn red policies, or assess the attacker's intrusion levels. To address this, we propose a Policy Learning Technique using imitation learning to learn policies for partially observable RL agents with discrete states and discrete actions. We apply this technique in an autonomous cyber environment to predict red agent's actions from network observations and defender actions. Integrated with a neurosymbolic cyber-defense agent, our method effectively handles different red policies and achieves high prediction accuracy across diverse simulated scenarios.

Learned representations in intelligent sensing systems are often evaluated by reconstruction fidelity or downstream prediction accuracy, but these criteria do not specify which latent distinctions are justified by the sensing process. In sensor-conditioned environments, nuisance factors can change measurements without changing the scene, while distinct scenes may be indistinguishable under limited sensing capability. This paper formulates sensor-conditioned representation correctness as preserving sensing-supported scene distinctions while suppressing nuisance-induced and sensor-unsupported variation. We introduce the scene-relevant observation quotient, a representation target induced by sensing-supported distinguishability after nuisance canonicalization, and develop Observation-Quotient Tucker-Structured Autoencoding (OQ-TSAE), a scene-nuisance factorized framework with diagnostics for false distinction, false merge, nuisance sensitivity, and latent ordering consistency. Experiments on a controlled benchmark show that quotient-consistent supervision improves representation-correctness diagnostics over reconstruction-oriented, metric-learning, and contrastive-learning baselines. Sensitivity, perturbation, and ablation studies show the importance of quotient-aligned supervision, reliable quotient relations, and quotient geometry. Complementary real-radar experiments show that a reconstruction-only OQ-TSAE variant retains competitive downstream utility, robustness under observation degradation, and low seed-to-seed variability. These results suggest that sensor-conditioned representations should be evaluated not only by predictive utility, but also by whether their latent geometry preserves sensing-justified scene distinctions.