Action
-
Robot Learning with Vision-Language-Action Models for Agricultural Manipulation
2026-10-25 16:26:18Hamid Hafezi, Javad Mohammadpour Velni, Mohammadreza DavoodiAbstract
Robotic manipulation in agricultural environments remains challenging due to variations in plant geometry, occlusion, and camera viewpoints. Pretrained vision-language-action (VLA) and generalist robot policies offer a promising approach to addressing these challenges by learning manipulation behaviors from multimodal observations and language instructions. However, adapting these policies to agricultural tasks requires domain-specific demonstrations, which are costly and time-consuming to collect on physical robots. In this work, we develop a simulation-to-real robot-learning framework for agricultural reaching and grasping that integrates task-specific demonstration collection and pretrained policy adaptation. A MuJoCo-based agricultural environment is developed with diverse tomato plant structures to support demonstration collection and controlled evaluation. Expert demonstrations are collected through joystick teleoperation, with a low-level controller used to execute robot actions. Two pretrained generalist robot policies, Octo and $\pi_{0.5}$, are adapted using simulated demonstrations, followed by further fine-tuning on a physical Franka Research 3 manipulator using real-robot demonstrations. The adapted policies are evaluated under in-distribution, viewpoint/spatial-shift, and out-of-distribution conditions in simulation, followed by physical-robot experiments. The influence of proprioceptive information on policy performance is also investigated. Results show that $\pi_{0.5}$ achieves higher task-performance scores than Octo and exhibits less degradation under distribution shifts. These findings demonstrate the potential of integrating pretrained generalist policies with task-specific demonstration learning for agricultural manipulation.
Abstract (translated)
URL
https://arxiv.org/abs/2610.26717
PDF
-
A Domain-Aware Pipeline for Structured Extraction and Retrieval-Augmented Question Answering over arXiv Superconductivity Literature
2026-10-25 15:48:48Ivan Iorsh, Aleksandr Kapitonov, Ivan Filianin, Matvey Lesonen, Nikita ChizhovAbstract
Scientific literature in materials physics is growing faster than researchers can manually inspect it, yet most retrieval tools rely on keyword matching or generic question answering that ignores domain-specific structure in PDF files. This paper presents an end-to-end pipeline that transforms arXiv superconductivity publications into a searchable corpus with structured metadata and retrieval-augmented question answering. The system combines thematic corpus construction, domain-aware extraction of materials and superconducting transition temperatures, and hybrid retrieval that merges dense chunk embeddings with structured publication fields. We test whether normalized domain attributes improve evidence retrieval and answer groundedness relative to lexical-only and dense text-only methods. Extraction is evaluated on a manually verified gold set of 59 superconducting materials, 46 of which have confirmed experimental critical temperatures. On the 33-material primary positive subset, domain-aware pair accuracy reaches 45.5% versus 9.1% for a regex baseline. Retrieval and generation are compared under a fixed language model (qwen2.5:7b-instruct) on a 100-question benchmark, varying only the retrieval mechanism: BM25, dense, hybrid, and hybrid with domain metadata. BM25 gives the highest source recall (Recall@5 of 47%), and adding extracted metadata to hybrid retrieval does not improve it (39% versus 41%), which we attribute to noise in the extracted fields; the main bottleneck is locating the supporting span inside a correctly retrieved paper. As a materials-informatics application, literature-derived temperatures are matched to an independent theoretical reference set: 200 materials appear in both sources, and 20 have a median experimental critical temperature above the theoretical value.
Abstract (translated)
URL
https://arxiv.org/abs/2610.34391
PDF
-
Outcome-Aware Agents: When Does Causal Structure Help a Forward Model Foresee the Consequences of Actions?
Abstract
Agents that act on real systems should predict what their actions will cause before acting. We study when causal structure helps a learned forward model do this, using a planner that rejects candidate actions whose simulated multi-step outcome is unsafe, in synthetic structural causal models where the true consequences are known. We compare the Action-Conditioned Latent Structural Causal Model (AC-LSCM: per-factor mechanisms over a learned sparse graph, actions as do-interventions) with three unstructured forward models on seven random graphs with 10 or 20 variables, a task configured by a fixed rule, and two controls: a do-wrapper that gives every model the same action semantics, and counterfactual data given to all models or to none. When its factors are the observed variables, AC-LSCM recovers the true graph exactly on every 10-variable graph and has lower three-step prediction error than every baseline in all 396 paired comparisons. Its excess safety violations over an oracle are 0.25 of the no-model level on 10-variable graphs (best baseline 0.31) and 0.41 on 20-variable graphs (0.53), and its margin is largest for two-variable actions never seen in training (0.18 versus 0.28). With learned latent factors the same model is worse than the baselines; the choice of factors matters more than the explicit do-operator or counterfactual data, and every planner stays far from the oracle. This version corrects an earlier one, whose model learned an empty graph and whose headline 36x safety improvement came from a task that tested only extrapolation of action values; we withdraw that claim. Code and all per-seed results are released.
Abstract (translated)
URL
https://arxiv.org/abs/2610.34385
PDF
-
AndroidFlux: Evaluating and Reward Modeling Failure Recovery Capabilities in Mobile Use Multimodal Agents
2026-10-25 15:39:24Gyubeum Lim, Yemo Koo, Ghunwoo Jang, Zsolt KiraAbstract
Mobile-use multimodal agents are typically evaluated by task success from a clean initial state, which does not isolate their ability to recover from intermediate errors. Because mobile environments change dynamically, a general mobile-use agent must complete tasks robustly even from unexpected states and in unseen environments. This requires knowing when and where an agent makes errors, and whether it can recover from errors and subsequent states. To evaluate this capability, we introduce AndroidFlux, a failure-recovery benchmark built on AndroidWorld. AndroidFlux restores recorded trajectories to controlled checkpoints before, at, and after the first recoverable error, so that different agents continue from the same device state and history. On these checkpoints, a complementary reward-modeling benchmark evaluates recovery-action preferences using downstream rollout outcomes. We further release an offline reinforcement learning (RL) dataset that improves both error recovery and overall task success. Our evaluations show that recovery performance only partially correlates with clean-start success, and that explicit reasoning helps only models trained with GUI-domain specific RL.
Abstract (translated)
URL
https://arxiv.org/abs/2610.34384
PDF
-
ThinkLocoManip: A Whole-Body Control Agentic Harness for Legged Loco-Manipulation
2026-10-25 15:23:26Nuo Chen, Shizeng Zhang, Yujie Yang, Jinghong Wang, Zhenyang Qu, Xiao-Ming Wu, Yunfei Guo, Zhongwen Lu, Haojie Liu, Chunpeng Lu, Qiao Sun, Xueyi Zhang, Jiangwei ZhongAbstract
In this paper, we propose ThinkLocoManip, an agentic harness for legged loco-manipulation designed to tackle complex, long-horizon tasks in large-scale environments. Existing approaches typically follow either direct action generation or a plan-then-act paradigm, both of which can struggle with the generalization and error-recovery capabilities required for extended task execution. In contrast, ThinkLocoManip adopts an interleaved reasoning-and-acting paradigm, allowing the agent to continuously reason about its current state, execute actions, observe their outcomes, and adapt its strategy accordingly. Building on this paradigm, our agentic harness orchestrates a suite of specialized tools for perception, navigation, manipulation, and locomotion, enabling flexible composition of heterogeneous capabilities. Unlike existing robotic agentic harnesses, ThinkLocoManip explicitly incorporates locomotion as a tool domain for agent reasoning and execution. We further design a set of skills that explicitly account for both tool usage within each functional domain and coordination across domains, supporting effective execution of complex behaviors over long horizons. To tackle task failure during long-horizon interactions, we develop keyframe thinking mechanism to preserve critical execution context and support efficient error recovery. Extensive real-world experiments demonstrate that ThinkLocoManip substantially outperforms representative methods based on both direct action generation and plan-then-act paradigms across challenging long-horizon legged loco-manipulation tasks. Notably, ThinkLocoManip achieves a success rate of 77%, more than twice that of the strongest plan-then-act baseline (37.1%). Our experiments further validate the effectiveness of interleaved reasoning and acting, tool orchestration, and coordinated skill execution.
Abstract (translated)
URL
https://arxiv.org/abs/2610.34370
PDF
-
SEEK: Skill Extraction with Evolving Knowledge against Frontier Science Agents
2026-10-25 15:19:13Cheng Gui, Xiang Zheng, Xingjun Ma, Kaidi Xu, Cong Wang, Weiying MaAbstract
Science agents increasingly package domain expertise as reusable skills that plan analyses, select tools, and execute workflows. This modularity creates a confidentiality risk: the interactions that deliver a skill's capability can also reveal the skill itself. We build an extensible benchmark of 40 protected skills across eight open- and closed-source science-agent platforms, comparing direct, fixed-strategy, and bandit-guided iterative probing under reference-based and reference-free evaluation. Leakage is pervasive: all eight platforms (100%) leak at least one protected skill, and four reach 100% attack success under a single direct request. To stress-test platforms that resist simple probing, we propose SEEK (Skill Extraction with Evolving Knowledge), a black-box, self-evolving red-teaming agent that consolidates response-derived evidence in persistent memory and evolves reusable attack skills, using only user-visible responses. On 12 skills from SpatialAgent and Biomni, SEEK achieves 100% attack success while reducing the mean first-success round by 31.8% and 38.2% relative to bandit-guided probing. Against a confidentiality policy that blocks all three benchmark protocols, SEEK still extracts one of six protected skills. These findings show that current science agents do not reliably protect their skills and that defenses must be evaluated against adaptive, multi-turn attackers.
Abstract (translated)
URL
https://arxiv.org/abs/2610.34365
PDF
-
EAST: Embodied AI Safety Testing via Systematic Fault Injection in Digital Twins
2026-10-25 15:10:51Yuhao Tian, Tao Li, Ling Zhou, Dave Towey, Zhaoxin Fan, Wenjun Wu, Rubing Huang, Tsong Yueh ChenAbstract
Task success is not a sufficient measure of safety for embodied artificial intelligence (AI), because robots may complete objectives while violating safety requirements. Existing evaluations often emphasize performance and robustness, but lack a systematic route from documented safety hazards to executable, verifiable tests. We present EAST (Embodied AI Safety Testing), a digital-twin framework that converts literature-grounded safety evidence into fault-injection tests across perception, instructions, environments, actions, and timing. EAST combines an expert-reviewed safety-fault taxonomy and eight executable operators with applicability and validity constraints. Its three-stage workflow generates test specifications, verifies fault realization during simulated execution, and independently assesses safety challenges, violations, task outcomes, and behavioral signatures using simulator-state evidence. The framework preserves execution traces and replayable records, supporting attribution and reproducibility across experiments. We evaluate EAST on six RoboTwin manipulation tasks and two policy architectures, ACT and RDT, against alternative testing strategies under matched budgets. Across five comparative campaigns, EAST targets all 74 admissible task-policy-operator combinations and obtains valid evidence for 63, while achieving safety-challenge exposure rates close to Random-Legal and Rule-Based. Moreover, 82 valid EAST executions complete their tasks despite safety violations, revealing risks concealed by success-oriented evaluation. These findings support systematic, validity-aware, reproducible safety testing beyond task-completion metrics.
Abstract (translated)
URL
https://arxiv.org/abs/2610.34358
PDF
-
RynnVLA-Latent: Scaling Robotic Manipulation Foundation Models with 3D-Grounded Latent Actions
2026-10-25 15:05:32Haoyu Zhao, Xingyue Zhao, Zhian Su, Bohan Hou, Yinzhou Tang, Fengyi Wang, Kehan Li, Yuming Jiang, Siteng Huang, Xin Li, Zhenyu Zeng, Mingxiu Chen, Deli Zhao, Zhongyu LiAbstract
Scaling vision-language-action (VLA) models is constrained by scarce action-labeled demonstrations and incompatible control spaces across robot embodiments. We present RynnVLA-Latent, a framework that turns heterogeneous interaction videos into scalable supervision for robot policy pretraining. At its core, RynnLAM learns continuous, 3D-grounded latent actions from visual transitions without requiring native robot controls. By combining residual future-feature prediction, camera-compensated 3D-flow supervision, and camera-motion decoupling, RynnLAM encourages representations that capture control-relevant motion rather than appearance changes alone. We apply RynnLAM to 30.1K hours of human, robot, and simulation videos, producing 745.52M latent-action descriptors. RynnVLA-Latent then learns to predict latent-action chunks from visual observations and language through flow matching, before adapting its pretrained backbone and action expert to embodiment-specific robot actions. The resulting 4B policy achieves 98.8% success on LIBERO, 84.7% on LIBERO-Plus without benchmark-specific adaptation, 51.9% average success across five VLABench tracks, and 88.6% average success on the RoboTwin 2.0 bimanual benchmark. It also achieves the highest observed success rates among the compared methods on six real-world manipulation tasks across dexterous-hand and mobile-gripper platforms. Together with representation ablations and data-scaling results, these findings demonstrate the potential of latent-action supervision for scaling robot policy learning beyond native action-labeled data.
Abstract (translated)
URL
https://arxiv.org/abs/2610.34356
PDF
-
A Computationally Efficient Non-Interpolation Virtual Trellis Approach for Mammogram Gradient-based Patterns Analysis
2026-10-25 15:04:23Shahid Rasheed, Muhammad Rashid, Seifedine Kadry, Salman A. AlQahtaniAbstract
Feature analysis is a crucial step in medical image analysis, particularly in breast cancer classification. Traditional image analysis methods often struggle with optimal patterns along the diagonal axis, computational efficiency and feature buffer storage. Different classical operators require high-efficiency hardware for the complex operations, such as image convolution and gradient analysis, which are used for processing complex machine learning models for training and test operations. Here, in this research, we introduce a non-interpolation virtual trellis-based (NVHT) operator that utilizes a filter kernel over the same square image without transformation to a hexagonal trellis. This proposed method is very efficient in reducing computational and storage costs. The proposed operator is used for improved feature extraction efficiency and accuracy in disease classification. This method outperforms conventional approaches by leveraging a hexagonal pixel representation while maintaining high sensitivity in different directions with less storage.
Abstract (translated)
URL
https://arxiv.org/abs/2610.34352
PDF
-
Spatially-Aware Self-Organizing Graph Autoencoders via Localized Hebbian Learning
2026-10-25 14:57:51Robin Ko\c{c}, Hazal Mo\u{g}ultay \"Ozcan, Fato\c{s} T. Yarman VuralAbstract
Traditional autoencoders rely on static network architectures to learn latent representations. Recently we suggested a dynamic architecture inspired by Hebbian plasticity, enabling topological self-organization among convolutional feature maps (R. Koç et al., Proceedings of the SIU, 2026). Early implementations of this paradigm infer relationships among feature maps using global average pooling and model their interactions through scalar-valued connections. This approach falls short to fully exploit the underlying spatial information, embedded in input images. Furthermore, abrupt transitions between different loss functions and training dynamics can introduce representation shocks, leading to unstable feature learning. In this paper, we introduce a Spatially-Aware Self-Organizing Graph Autoencoder (S-LooM) to address these structural bottlenecks. By using spatial cross-correlation, and spatial kernels, S-LooM establishes dynamic connections among feature maps. We replace static scalar message passing of autoencoders with learnable spatial kernels and eliminate high-frequency deconvolutional artifacts observed in decoders using nearest-neighbor upsampling. Via a soft-blending training curriculum, the latent topology aligns smoothly during training. Our experiments show that S-LooM yields a more stable topology, compared to the state of the art dynamic architectures. Meaning, the graph structure that emerge has more resistance to feature entanglement and noise. It forms functional communities responsible for visual attributes, such as illumination, shape and texture even under high connection densities.
Abstract (translated)
URL
https://arxiv.org/abs/2610.34346
PDF
-
Xing USP: A Cross-Family Semantic Contract for Information Extraction and Intent-Slot Parsing
2026-10-25 14:56:09Zhongqiu Li, Boyang Wang, Ruiyu Fang, Shiquan Wang, Di Wu, Jiayi Liang, Feilong Zhang, Hongfeng Du, Xiaomeng Huang, Mengjiao Bao, Luwen Pu, Xuewei He, Yu Zhao, Chao Wang, Kaidong Yu, Bingkai Yang, Qingjun Zhu, Shaoyi Du, Shuangyong Song, Yongxiang LiAbstract
Structured semantic parsing is typically judged by whether the correct category is assigned, which leaves value recovery and instance-level associations unexamined: a system may recognize an event yet bind an argument to the wrong instance, or identify an intent yet supply an incorrect slot value. We present USP, a cross-family semantic contract that represents categories, fields, values, and instance associations through a common interface for named-entity recognition, relation extraction, event extraction, and intent-slot parsing. The contract is paired with five diagnostic matching views---category agreement, field recognition, field-value pairing, instance-bound assignment, and complete-instance recovery---that ask distinct questions about the same output. Using this contract, we normalize 61 datasets into a unified corpus with shared schemas, prompts, and matching rules, and compare six models on a fixed test subset. Our fine-tuned Xing4.0-29B-Intent attains 57.70 task-balanced Field-Value Micro-F1 (equally weighted across the four task families) and ranks first in every family and on all five views; its margin over the strongest alternative widens from 2.93 points for category matching to 20.73 points for complete-instance matching. This widening gap shows that category agreement alone understates structural deficits. USP thus offers the community a shared representation, a normalized 61-dataset corpus, and a diagnostic evaluation protocol that make structural agreement measurable and comparable across extraction and intent-slot tasks.
Abstract (translated)
URL
https://arxiv.org/abs/2610.34343
PDF
-
Developer Task Delegation to LLMs and the Trust Factor: A Mixed Methods Study
2026-10-25 14:28:10Samuel Ferino, Rashina Hoda, John Grundy, Christoph Treude, Hashini GunatilakeAbstract
With the rise of Large Language Models (LLMs), software developers are presented with a powerful new approach: delegating software development tasks to LLMs. Existing studies have explored task delegation for a subset of tasks and factors. Critically, what tasks developers delegate to LLMs, to what extent, and why they delegate them, is not yet well understood. We explore these questions through 22 interviews and a survey with 160 software practitioners. Developers reported employing four modes of task delegation to LLMs: full delegation, partial delegation, self-delegation, and task avoidance. Drawing on our qualitative findings, we discuss the rationales behind these choices and explore how task difficulty, task risk, developer motivation, and trust appear to be associated with software development task delegation to LLMs. Our statistical analysis reveals that trust is a significant factor associated with task delegation to LLMs. Finally, we share actionable implications for human-AI collaboration in software research, and for software practitioners to make the most of their newly found powers of delegation.
Abstract (translated)
URL
https://arxiv.org/abs/2610.34317
PDF
-
Matched Capabilities, Different Behaviors: A Controlled Study of MCP and CLI Interfaces for Language Agents
Abstract
Language agents increasingly act through external tools and interactive environments, making the interface between the model and its executable capabilities part of the decision process. Model Context Protocol (MCP) and command-line interfaces (CLI) can expose the same semantic operations through markedly different model-facing representations. We study whether these representational differences change agent behavior when executable capabilities are matched. We introduce Matched Interface Comparison and Analysis (MICA), which renders MCP and CLI from a shared semantic action contract and routes decoded actions through the same runtime. Across six models and three interactive environments, matched interfaces produce substantial differences in task performance and interaction cost. These effects extend beyond surface validity to semantic action selection, failure pathways, and recovery. Matched-state probes localize sensitivity to capability presentation, realization, and feedback. Paired natural trajectories reveal structured divergence in 94.3% of 2,004 episodes and show that early differences propagate through subsequent interaction. This analysis identifies interface realization as a direct intervention boundary. We therefore introduce Decoupled Action Realization and Targeted Repair (DART), which preserves validated semantic commitments while repairing interface-specific realization. DART improves benchmark score in all six evaluated settings by up to 9.0 points and reduces MCP token cost in all three environments by up to 27.1k tokens per task. Together, these results establish interface representation as part of the closed-loop decision environment shaping agent behavior.
Abstract (translated)
URL
https://arxiv.org/abs/2610.34298
PDF
-
AdaPolicy: Action-Driven Anticipatory Robot Policies via Next-Chunk Prediction
2026-10-25 13:23:22Fan Yang, Yuting Su, Chen Zhao, Fugui Fan, Peiguang JingAbstract
Endowing robot policies with anticipatory capabilities has proven effective in improving control performance. However, existing methods predominantly rely on explicit supervision derived from future observations, in either pixel or latent space, at the cost of additional architectural and computational complexity. Departing from explicit future supervision, we ask whether such foresight can be elicited directly from the action signal. To this end, we propose \textbf{AdaPolicy}, an \textbf{A}ction-\textbf{d}riven \textbf{a}nticipatory policy framework that shapes future-aware representations via cascaded next-chunk prediction. Specifically, an Online Action Head jointly produces the current action chunk and an evolved latent state. Building on this state, a cascaded Unrolled Action Head forecasts the next chunk, and the resulting loss flows back through the evolved state to instill implicit foresight into the online policy. Importantly, this training-only unroll requires no future-observation encoding, keeping AdaPolicy lightweight to train with zero inference-time overhead. Experiments on MetaWorld, RoboTwin 2.0, LIBERO, and real-world deployments demonstrate consistent gains across diverse settings, including up to $14$ percentage points on millimeter-precision tasks such as earphone assembly and shoelace threading. More details and visualizations are available at \href{this https URL}{this https URL}.
Abstract (translated)
URL
https://arxiv.org/abs/2610.34280
PDF
-
Recovering Business Rules as Cascade Lookups for Enterprise Field Prediction
2026-10-25 12:33:59Changxu DuanAbstract
Enterprise sales-order systems rely on manually maintained configuration rules to determine fields such as payment terms, shipping points, and sales ownership. When these rules are incomplete or stale, operators must enter values by hand, and downstream errors spread into billing, logistics, and reporting. This paper studies whether missing or outdated rules can be recovered from historical transactions instead of being relearned as opaque statistical patterns. We introduce a knowledge-graph-guided workflow in which an LLM agent proposes candidate business keys, meaning column combinations such as customer and document type, validates them through SQL profiling, and compiles the resulting evidence into deterministic cascade lookups. The LLM is used only during offline rule discovery. At inference time, prediction is a transparent key lookup with no model-serving requirement. Experiments on enterprise sales-order prediction and public relational benchmarks show that cascades are most effective when fields follow stable business rules, while learned graph models remain preferable when prediction depends on temporal or neighborhood patterns. Drift analysis further turns residual errors into diagnostics for configuration maintenance. Code and data: this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2610.34258
PDF
-
Variational Cognitive Memory Abstraction for Long-Horizon Vision-Language-Action Models
2026-10-25 12:24:57Xinyu Liu, Daiying Tian, Chenwei Tang, Ruyang Han, Chengyang Zhang, Qingyu Zeng, Wei Ju, Liangli Zhen, Jiancheng LvAbstract
Vision-Language-Action (VLA) models have shown remarkable success in robotic manipulation, yet their ability to execute long-horizon tasks remains limited by insufficient long- term memory modeling. Existing methods either rely on implicit temporal representations within policy networks, where historical information can be gradually diluted, or maintain explicit memory buffers that suffer from redundancy, noise accumulation, and inefficient retrieval. In this paper, we propose CogMemory, a novel variational cognitive memory framework that enables VLA systems to selectively preserve and reuse task-relevant historical information. Our approach encodes multimodal trajectories into compact latent memory representations through a selective state- space memory encoder, multi-slot temporal abstraction, and variational information compression. During action generation, dynamic cross-attention retrieval with gated fusion adaptively incorporates relevant memories into policy prediction. With merely 3.45M additional parameters, CogMemory stays lightweight and seamlessly fits both early- and intermediate-fusion VLA backbones. Extensive experiments on long-horizon robotic manipulation benchmarks demonstrate consistent improvements across diverse VLA backbones.
Abstract (translated)
URL
https://arxiv.org/abs/2610.34251
PDF
-
LiveOption: Evaluating LLM Agents in Structured Option Trading with Nonlinear Payoffs
2026-10-25 11:22:19Haochen Luo, Yifan Li, Binh Minh An, Xiaolong Luo, Zhengzhao Lai, Yuan Zhang, Chen LiuAbstract
Large language models (LLMs) and multi-agent systems (MAS) have shown promise in financial decision-making, yet existing evaluations focus on equity trading and primarily assess directional prediction, overlooking the structural complexity of derivative markets. Option trading introduces fundamentally different challenges, including nonlinear payoffs and multi-leg strategy construction, requiring structured decisions rather than simple directional bets. We introduce LiveOption, an evaluation framework for LLM-based agents in option trading. LiveOption formulates the problem as structured sequential decision-making under realistic execution and capital constraints, and provides a reproducible environment with standardized interaction protocols. The framework includes three task suites covering portfolio overlays, event-driven earnings trading, and 0DTE intraday trading. We further propose a hierarchical metric suite that evaluates action validity, decision quality, risk characteristics, and outcome-level performance. Experiments show that current agents often fail to achieve competitive returns in most scenarios. LiveOption offers a principled testbed for evaluating structured decision-making beyond outcome-based metrics. Project Website: this https URL
Abstract (translated)
URL
https://arxiv.org/abs/2609.33470
PDF
-
From Presence Labels to Bounding Boxes: Can Multimodal Large Language Models Scale Surgical Instrument Localization?
2026-10-25 11:18:07Mohamed Hamdy, Muraam Abdel-Ghani, Fatmaelzahraa Ahmed, Sifna Nasar, Mariam Ahmed, Khalid Al-Jalham, Abdulaziz Al-Ali, Shidin BalakrishnanAbstract
Spatial localization of surgical instruments through bounding-box annotations is a foundational component of surgical scene understanding, supporting detection, tracking, interaction analysis, and grounded vision-language applications. Yet existing annotations often cover selected frames, video subsets, or dataset variants rather than complete procedures. Manual extension is time-consuming and expert-dependent. Additionally, weakly supervised and detector-based pseudo-labels require task-specific training and may be imprecise. In this study, we design a scalable MLLM-based annotation framework that converts surgical frames and existing instrument-presence labels into accurate bounding boxes, and systematically analyze its reliability, cost, and design sensitivity. We instantiate the framework using six MLLMs and a multi-call configuration on laparoscopic cholecystectomy in Cholec80. Using a carefully designed stratification scheme, two annotators independently label 1,000 frames, and an expert surgeon reviews one annotation set to establish the reference. Results show that instrument-set F1 ranges from 88.2 to 92.7,whereas mIoU ranges from 0.180 to 0.735, showing that comparable recognition can conceal large localization differences. A structured multi-stage annotation pipeline achieves 0.784 mIoU, compared with 0.735 for the best single-call baseline and 0.814 human-human agreement. Using the best accuracy pipeline, we create and publicly release a spatial extension of Cholec80 comprising 285,977 bounding boxes across all 184,498 annotated frames from the 80 videos. The annotations and human references are made available on GitHub
Abstract (translated)
URL
https://arxiv.org/abs/2610.34218
PDF
-
MIS-VQA: An Annotation-Driven Surgical Vision-Language Dataset for Complex Surgical Understanding
2026-10-25 10:56:26Mohamed Hamdy, Iheb Zouari, Fatmaelzahraa Ahmed, Muraam Abdel-Ghani, Muhammed Arsalan, Ponnuthurai Suganthan, Khalid Al-Jalham, Abdulaziz Al-Ali, Shidin BalakrishnanAbstract
Modern surgical AI increasingly requires multimodal understanding rather than isolated perception, yet existing datasets remain largely limited to single-task supervision such as recognition or templated question answering. Current medical VQA resources often rely on synthetic templates, lack natural descriptive supervision, or provide sparse annotations that hinder reasoning about spatial relations and procedural context. We introduce MIS-VQA, a surgical vision-language dataset generated through an annotation-driven pipeline that converts heterogeneous surgical annotations into structured textual scene representations. Using EndoVis18 and CholecT50, we transform phase, instrument, interaction, triplet, and segmentation labels into unified contexts that guide large language models to generate clinically coherent question-answer pairs. MIS-VQA supports four response types-concise and detailed descriptions, perception-level grounding, and multi-step reasoning-enabling hierarchical evaluation from atomic visual grounding to compositional reasoning. To address annotation sparsity, we incorporate confidence-filtered segmentation and cross-annotation consistency checks, enabling spatially grounded supervision at scale. MIS-VQA establishes a unified benchmark for surgical vision-language understanding with spatial supervision and multi-tier reasoning evaluation. To demonstrate its utility, we introduce LLaVA-SurgXpert, a family of surgical vision-language models with visual encoders to analyze how visual representations influence multimodal reasoning. Experiments show consistent improvements over existing surgical VQA baselines while revealing capability gaps between perceptual grounding and higher-level reasoning hidden in prior benchmarks.
Abstract (translated)
URL
https://arxiv.org/abs/2610.34210
PDF
-
CORMA: Constraint-Routed Decision Mamba for Generalizable Multi-Constraint Offline Safe Reinforcement Learning
2026-10-25 10:54:34Changxu Ni, Feiran You, Yongan Guo, Hongyang Du, Kang Wei, Le Liang, Yiyang Ni, Jun LiAbstract
Offline safe reinforcement learning (RL) learns safety-aware policies from static datasets while satisfying safety constraints during deployment. Motivated by the strong trajectorymodeling capabilities of Transformers, recent Decision Transformer (DT)-based methods formulate offline safe RL as conditional sequence modeling by conditioning action generation on return-to-go (RTG) and cost-to-go (CTG) tokens, thereby enabling adaptation to varying safety budgets. However, existing DT-based methods are developed for single-constraint tasks. For tasks with multiple constraints, generalizing learned policies to unseen combinations of safety budgets remains challenging due to the conflicting requirements imposed by heterogeneous constraints. Moreover, the quadratic complexity of Transformer architectures significantly increases the computational overhead in multi-constraint tasks, as representing additional constraints requires more CTG tokens. Recently, Mamba has emerged as an efficient sequence-modeling backbone, offering linear computational complexity and strong generalization capabilities. In this paper, we propose CORMA, a constraint-routed decision Mamba framework for generalizable multi-constraint offline safe RL. CORMA decomposes policy inference into constraint-aware RTG estimation and adaptive action generation, where constraint-specific experts model feasible RTG distributions and a router-guided product-of-experts mechanism fuses these distributions to provide safety-aware guidance for a Mamba-based action decoder. A three-stage optimization strategy is designed to improve training stability. Extensive experiments on OSRL benchmarks demonstrate that CORMA achieves superior reward-safety trade-offs, reduces inference latency by 37% compared with DT-based methods, and exhibits stronger zero-shot generalization to unseen safety-budget combinations than offline safe RL baselines.
Abstract (translated)
URL
https://arxiv.org/abs/2610.34208
PDF
