Sparse
-
The Role of Radiometric Features in Cross-Site Leaf-Wood Segmentation of LiDAR Point Clouds
2026-09-18 15:31:39Roman Kaharlytskyi, Derek T. Robinson, Roberto GuglielmiAbstract
Leaf-wood segmentation of individual trees from LiDAR point clouds is essential for quantitative structure models (QSMs) used in non-destructive biomass estimation. Existing segmentation methods typically exclude radiometric features (e.g., intensity, return number) to maximize cross-sensor compatibility. We challenge this design choice by evaluating cross-site and cross-platform generalization: training on the public Heidelberg dataset (terrestrial TLS, 1550nm) and testing on a novel dataset from Ontario, Canada (RPA-LS, 905nm). Results show that geometry-only methods - including state-of-the-art deep learning models trained on high-density LiDAR datasets - fail to generalize to the sparse, top-down geometry of aerial scans, achieving F1 scores <= 0.56. Incorporating radiometric features (intensity, return number, number of returns) improves F1 to 0.61, but more critically, increases wood recall by 119% from 0.16 to 0.35. Furthermore, geometry-only approaches often result in fragmented stem and branch components. We find that leveraging radiometric features preserves greater structural connectivity, resulting in more coherent architectures that are better suited for QSM reconstruction. We demonstrate that while geometric patterns are view-dependent and prone to overfitting scan patterns, radiometric features encode physical material properties that generalize across disparate sensors and environments.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21903
PDF
-
RheoSampling: Resolving the One-Hot Dilemma in Stochastic Dynamic-Tree Speculative Decoding
2026-09-18 14:28:28Qiao Hu, Yepeng Weng, Bo Zhang, Takehisa YairiAbstract
Speculative decoding accelerates LLM inference by drafting multiple tokens in parallel, with tree-based methods further improving efficiency through hierarchical structures. Dynamic-tree methods such as EAGLE-3 perform well under greedy decoding via deterministic top-K expansion and global pruning. However, in stochastic decoding (T>0), this mechanism collapses the draft distribution into one-hot probabilities, causing a severe drop in acceptance rate. This creates a dilemma: dynamic-tree methods sacrifice stochastic sampling to preserve context-aware topology, while static-tree methods preserve stochastic sampling with context-agnostic structures. The issue arises because the same probability distribution is used for two conflicting tasks: constructing the tree and verifying tokens. This coupling makes direct injection of randomness challenging due to the resulting stochastic process. We resolve this by decoupling these roles: RheoSampling assigns a token sampled from the draft distribution a proxy probability for tree expansion and pruning alongside its true sampling probability for verification. Specifically, we inject a sampled token among the deterministic top-K slots and treat it with different probabilities during construction and verification, making RheoSampling the first dynamic-tree method with both context-aware top-K construction and stochastic sampling while maintaining losslessness. We establish the lossless guarantee through an equivalence-class analysis that compresses the stochastic tree space into tractable classes. An OT-based verification strategy and a sparse draft mechanism ensure that theoretical gains translate into practical efficiency. Experiments across LLMs and benchmarks demonstrate improvements in acceptance rate and speedup over state-of-the-art dynamic tree methods. This framework may provide a template for analyzing stochastic tree structures.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21827
PDF
-
VideoReloc: Long-Term Indoor Video Relocalization against a Kilobyte-Scale Semantic Scene Graph
2026-09-18 14:16:41Qianru Li, Xuyang Chen, Xuqin Wang, Zhenghao Zhang, Hongyi Luo, Tao Wu, Daniel Cremers, Lu Liu, Yanfeng ZhangAbstract
Given a compact semantic scene graph, long-term indoor video relocalization estimates a map-frame trajectory after lighting and furniture changes. Visual methods rely on appearance and become unreliable under these changes; localizing one frame at a time from object classes and geometry instead leaves sparse, ambiguous evidence. We introduce VideoReloc, whose adaptive clips use odometry to gather spatial evidence until object and motion criteria are met, adapting query length to the observed scene. Its run-level decision rechecks conflicting placements using evidence accumulated across connected clips, stabilizing the trajectory beyond adjacent-clip tracking. Hypothesis-first registration proposes poses from object triplets and verifies each using clip-wide object centers and box surfaces. Orientation-aware refinement uses box faces, gravity and wall directions to resolve ambiguity in camera orientation and refine the full pose. This reframes sparse-map relocalization as verification of spatially extended video queries, moving discriminative support from stored appearance to temporal context and permitting a 100 kB map of class-labelled boxes. On RIO10 and ReplicaCAD, the all-frame localization success rate at 1 m/10$^\circ$ is 73.5% and 61.1% under causal evaluation, rising to 90.6% and 74.8% with clip closure. The evaluated per-frame scene coordinate regressors reach up to 47.6% and 49.8%, respectively, with maps of 12.6-42 MB. Project page: this https URL
Abstract (translated)
URL
https://arxiv.org/abs/2609.21804
PDF
-
From Pretraining to Proficiency: Real-World Subtask RL for Long-Horizon Manipulation with Minimal Human Intervention
2026-09-18 14:03:49Sichang Su, Benjamin Yang, Zhiyun Deng, Boyuan Liang, Yip Fun Yeung, Zelin Wang, Lingfeng SunAbstract
A pretrained robot foundation policy may execute most of a long-horizon task yet repeatedly fail at a few critical subtasks. Collecting additional full-task demonstrations for supervised fine-tuning (SFT) requires operators to repeat behaviors the policy already performs well. Reinforcement learning (RL) fine-tuning offers a promising path to bridge this gap, but existing approaches struggle to solve long-horizon tasks using only sparse rewards. We present PARTS (Policy Adaptation with RL on Targeted Subtasks), a real-world subtask RL framework that concentrates practice at these bottlenecks while allowing training rollouts to proceed with minimal human intervention. The frozen pretrained policy supplies nominal actions throughout execution, while agent-generated selectors and success verifiers activate residual corrections and provide local outcome rewards. These rewards support learning from successful subtasks even when complete-task successes are scarce. Training combines online RL with success-reweighted retraining, and each retrained residual policy is redeployed to collect further experience. Humans identify bottlenecks during setup and perform physical resets when needed. On bimanual YAM and single-arm Franka tasks, PARTS improves complete-task success from 32% to 61% and from 50% to 95%, respectively, using tens of minutes of real-world RL rollouts per task on average. Compared with existing real-world RL fine-tuning methods, PARTS raises full-task success by more than 25% under the same robot-rollout budget while requiring less human involvement.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21788
PDF
-
SynthDemo-RL: Breaking the Zero-Reward Barrier in VLA Adaptation with LLM-Guided Synthetic Demonstrations
2026-09-18 11:48:05Hiroaki Kingetsu, Hiroaki Kurihara, Kaoru Yokoo, Kenji Fukumizu, Manohar KaulAbstract
Fine-tuning Vision-Language-Action (VLA) models commonly relies on human teleoperation demonstrations, while reinforcement learning (RL) with sparse binary rewards faces an exploration challenge when successful trajectories are rarely sampled. We propose SynthDemo-RL, a teacher-student framework in which an automated teacher converts simulator-privileged state into successful manipulation trajectories, a VLA student is distilled from them by supervised fine-tuning (SFT), and PPO with binary task-success rewards refines the student. We study reward coverage, the fraction of tasks for which at least one success is observed under the fixed evaluation protocol, as a complement to the average success rate. On LIBERO-PRO, a public benchmark of perturbed LIBERO tasks for which no demonstrations exist, 27 of 57 scored tasks are at exactly 0% success for a pi_0.5 policy fine-tuned on the original LIBERO tasks. Direct PPO from this policy, under the same PPO recipe and the same RL compute as SynthDemo-RL's refinement stage, rescues 10 of these 27 tasks and leaves 17 at 0%. SynthDemo-RL, with 50 synthesized trajectories per task and no new human demonstrations, rescues all 27 and reaches average success rates of 97.8% and 97.1% on the Position and Task axes of LIBERO-PRO, respectively. On standard LIBERO, the same pipeline reaches 96.0% with no human demonstrations, within 1.7 points of pi_0.5 trained on 50 human demonstrations per task. We further validate the pipeline on RoboTwin 2.0 and verify that trajectories from a policy trained in a MuJoCo twin execute open-loop on a physical robot.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21650
PDF
-
HAT: Hypothesis-Anchored Tracking for Video Monocular Spacecraft Pose Estimation
2026-09-18 10:28:30Andr\'e Lopo, Atabak Dehban, Rodrigo VenturaAbstract
Monocular 6-DoF pose estimation of non-cooperative targets is important for on-orbit servicing and debris removal. A single-image estimator can confuse near-symmetric spacecraft orientations, and tracking can preserve an incorrect pose. We present Hypothesis-Anchored Tracking (HAT), a causal framework that uses inter-frame motion to select among competing CAD-based pose hypotheses before alignment and fusion. Rather than independently choosing the highest-scoring hypothesis in each image, HAT retains competing orientation histories and selects a pose to anchor the relative trajectory estimated by monocular SLAM. Sparse anchors and pose fusion provide per-frame estimates after initialization without revising past outputs. The method requires only a calibrated RGB sequence, a metric CAD model, and target image regions, which can be supplied by detection or segmentation. The pretrained pose and SLAM networks require no target-specific training or fine-tuning. We evaluate two versions, Mega-HAT and Pico-HAT, using MegaPose and PicoPose, on SPARK-2024, SwissCube and SHIRT, with YCB-Video assessing performance outside the space domain. Using one temporal configuration per method, the arithmetic means of the four dataset-wise comparisons show 9.4% lower mean pose error and 3.76 times the sustained input FPS for Mega-HAT relative to independent MegaPose, and 23.9% lower mean pose error and 2.42 times the FPS for Pico-HAT relative to independent PicoPose. Mega-HAT ablations on SPARK and an offline reference examine component contributions and the effect of revising past estimates.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21597
PDF
-
SABER: Learning Attention-based Semantic Affordance for Legged Locomotion
2026-09-18 10:02:36Hari Prasanth Palanivelu (Michael), Samuel Sze (Michael), Kennard Garrison Johannes (Michael), Albertus Hendrawan Adiwahono (Michael), Meng Yee (Michael), ChuahAbstract
Perceptive legged locomotion has advanced rapidly by integrating terrain geometry into learned policies, yet the integration of terrain meaning remains sparse: a pipe, a patch of grass, or a fragile box may be geometrically traversable while being inappropriate for contact. In industrial environments, where legged robots increasingly operate, a single misplaced step can damage fragile equipment, destabilize the robot, or endanger the site. To address this, we introduce SABER, a planner-free reinforcement-learning policy that jointly reasons about terrain geometry and semantic contact permission. The policy consumes a unified terrain-affordance map, where each cell encodes local 3D geometry and a semantic contact cost. We augment cross-attention with a learned, signed semantic bias: an additive term on the attention logits, gated by the contact cost, that reweights flagged cells by their distance from the nearest foot. A hazard therefore reshapes attention where it can still affect the next foothold, and its influence fades where it cannot. The resulting policy selects footholds on permitted support and keeps the leg clear of forbidden regions throughout the swing phase. We perform a systematic ablation that isolates the contribution of each architectural component; removing the semantic bias alone increases forbidden contacts by 55% while velocity tracking is unchanged. We validate the policy on a Unitree B2, demonstrating sim-to-real semantic contact selection across indoor and outdoor environments and four semantic obstacle classes.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21572
PDF
-
OneBid: A Unified Auto-Bidding Foundation Model for Diverse oCPX Advertising Scenarios
2026-09-18 09:43:09Yewen Li, Peng Jiang, Yitian Li, Pengfei Lv, Xialong Liu, Peng Jiang, Qingpeng CaiAbstract
Auto-bidding is central to computational advertising, where strategies must maximize advertisers' conversion value under economic constraints. It has evolved from rule-based controllers to reinforcement learning and generative methods such as Decision Transformer (DT). Yet these methods increasingly mismatch the prevailing optimized cost-per-X (oCPX) paradigm, which spans heterogeneous scenarios (e.g., registration, purchase), each served by a separate model, leading to fragmented pipelines and underexploring cross-scenario modeling. Inspired by foundation models like LLMs, unifying these oCPX scenarios into one model raises three challenges: multi-objective control, scalable capacity under strict latency, and safe offline policy improvement. We present OneBid, a unified auto-bidding foundation model that learns a reusable backbone from heterogeneous oCPX logs and adapts it to scenario-specific deployments via offline post-training. Building on DT, OneBid extends single Return-to-Go conditioning to two atomic signals, Return-to-Go for conversion value and Cost-to-Go for cost ratio, plus value-aware regularization on next-action prediction. To absorb distributional heterogeneity, we design a sequence-level Mixture-of-Experts architecture, where shared experts encode cross-scenario knowledge and sparsely-routed experts capture scenario-specific patterns at low latency, yielding consistent scaling with model size and data. During post-training, we align the backbone with scenario preferences via Critic-guided Relative Offline Policy optimization (CROP): a learned critic scores candidate actions group-relatively, avoiding the unsafe online exploration of GRPO-style fine-tuning while constraining policy shift to reduce OOD risk. Validated via online A/B tests and fully deployed at Kuaishou, OneBid delivers an overall +2.2% ADVV gain on oCPX Ads, peaking at +13.1% in the ROAS scenario.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21550
PDF
-
From Retrieval to Recognition:How Vision--Language Models Become OCR Specialists
2026-09-18 09:32:37Yuanxiang Huangfu, Hanmeng Zhong, Linqing Chen, Jeffrey Tiong Jee HuiAbstract
Does a general vision--language model acquire specialized OCR ability by developing a new reading circuit or by reusing an existing mechanism? We address this question in the setting of full-sequence OCR, rather than local-answer retrieval. Using an evidence-grounded protocol with held-out causal interventions, we identify sparse and stable OCR-head sets in GLM-OCR, MinerU2.5, and PaddleOCR-VL-1.6. We then investigate the mechanistic origin of these OCR heads by comparing them with independently identified textual retrieval/copy heads in general VLMs. Across two general VLMs, visual OCR heads strongly overlap independently identified textual retrieval/copy heads, yielding untuned top-20 intersections of 73.3% and all-head Spearman correlations of 0.677-0.886. The overlap and causal interventions suggest that full-sequence OCR operates as dense sequential multimodal copy-and-paste, repeatedly retrieving visual evidence and routing it to the current output position. Finally, we examine how this shared circuit changes as a general VLM becomes an OCR specialist. Matched base-to-specialized comparisons show that OCR specialization largely preserves head identity, retaining 17-20 of the top 20 heads per task with all-head rank correlations of 0.874-0.942, while redistributing their functional and causal strengths.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21543
PDF
-
Driving on Registers, Reasoning on Risk: Risk-Aware Occupancy for Register-Based End-to-End Autonomous Driving
2026-09-18 08:39:36Jiaxing Chen, Hengduo Zou, YuKai Qin, Yiren Zhao, Lidong Yu, Bolin GaoAbstract
Multimodal trajectory prediction improves behavioral coverage in end-to-end autonomous driving, but existing methods remain limited by sparse scene representations. Incomplete evidence leads to low-quality candidate generation and unreliable ranking among geometrically similar trajectories. On a register-based baseline, bad and poor candidates constitute 19.74% of the candidate set, while the oracle-best candidate ranks only 33.9th on average. We propose RRDrive, which introduces risk-aware occupancy as a dense, temporally aligned, and trajectory-queryable representation. Its global structure guides high-quality multimodal generation, while candidate-conditioned risk queries support fine-grained selection. We further construct RiskOcc4D-NAVSIM with automatic risk annotations. RRDrive achieves a selected-trajectory PDMS of 0.951, representing a 1.5% relative improvement over the baseline (0.937), and improves the average candidate PDMS by 7.7%. In challenging scenes, it improves candidate PDMS by 30.2% and increases the Spearman correlation among good candidates by 0.41, from 0.26 to 0.67. To move beyond this oracle setting, we further develop an external RiskOcc predictor, a perception module that estimates risk-aware occupancy directly from sensor inputs. The competitive performance validates the representation's feasibility.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21486
PDF
-
Risk-Aware Occupancy for Safety-Oriented End-to-End Autonomous Driving
2026-09-18 08:18:07Jiaxing Chen, Hengduo Zou, Yiren Zhao, Bolin GaoAbstract
Sparse representation formulates the environment perception for the end-to-end driving system as a set of discrete elements like objects and lane lines. This formulation meets safety risks in crowded, occluded scenes dealing with unstructured obstacles, uncertain regions, and intricate interactions. In this paper, we propose a dense representation, risk-aware occupancy, to characterize planning-relevant risks in an explicit and uniform manner. It jointly encodes global scene occupancy, map-derived traffic constraints, and future dynamic agent occupancy into a unified BEV map. The unified BEV map captures the risk evidence for trajectory planning in both spatial and temporal dimensions. We design an E2E network, ROIDrive, to realize risk-aware occupancy. It predicts risk-aware occupancy with an independent branch and injects it into planning queries for safety-oriented trajectory generation. In addition, to quantify the safety problem, we introduce RiskOcc4D-nuScenes built upon nuscenes and occ3d-nuscenes. Our risk-aware occupancy yields relative open-loop collision reductions of 52.9% under the UniAD metric and 35.0% under the ST-P3 metric on nuScenes.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21470
PDF
-
FootQuery: Future-Touchdown-Guided Retrieval from Depth History for Perceptive Humanoid Locomotion
2026-09-18 07:59:24Tao Dong, Jia Yu, Yuxuan Fan, Linna Zhao, Jiaqi Gong, Andong Yang, Chao Gao, Guyue ZhouAbstract
Humanoid locomotion over complex terrain requires anticipating footholds that may no longer be visible at touchdown. Limited camera coverage and self-occlusion make it necessary to retrieve relevant terrain information from earlier observations. We present FootQuery, a perceptive locomotion framework that queries depth history using each foot's predicted next touchdown. The policy predicts touchdown locations and uncertainty from proprioception and uses these distributions, together with per-foot features, to query sparsely sampled historical depth frames. During training, realized contacts are projected into historical images to supervise retrieval at the regions where those contacts were visible. The retrieved per-foot features are fused with global visual memory to generate control actions. A progressive force-assistance curriculum supports early exploration, while event-consistent tread-midline shaping encourages coordinated stair contacts. Deployment requires only proprioception and onboard depth images. In simulation, the complete framework outperforms its component ablations on the most challenging tested stairs, gaps, and platforms. Real-world experiments on a Unitree G1 demonstrate continuous traversal with a single policy across outdoor stairs and indoor routes combining stair ascent and descent, platforms, and gaps. These results support organizing visual history around anticipated contacts for perceptive humanoid locomotion.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21447
PDF
-
Interference-Driven Clustered Optimisation for FM Spectrum Coordination
2026-09-18 07:53:42Federica Mangiatordi, Emiliano PallottiAbstract
Cross-border FM spectrum coordination involves protecting foreign broadcasting services while preserving domestic coverage, amid increasingly large radio-planning datasets containing thousands of transmitters and millions of transmitter-pixel relationships. In such scenarios, conventional optimisation approaches become computationally demanding due to the high dimensionality of the associated power-control problem. This paper proposes an interference-driven clustered optimisation framework for large-scale FM spectrum coordination. The proposed method exploits the observation that violations of foreign-service protection are typically dominated by a limited subset of transmitters. Protected services are therefore analysed to identify dominant interferers and quantify their impact on interference. These relationships are represented through an interference graph from which optimisation-oriented transmitter clusters are extracted. The clusters decompose the global power-control problem into smaller optimisation tasks solved with clustered simulated annealing, followed by a global refinement that captures residual inter-cluster interactions. Coverage and interference are evaluated using frequency-dependent protection criteria and a dynamic strongest-service assignment model. To enable operational-scale planning, the framework uses sparse matrices and GPU-accelerated computations. Tests on realistic cross-border FM coordination scenarios show that the clustering strategy greatly reduces optimisation complexity and runtime while maintaining foreign-service protection and domestic coverage. The method also yields an interpretable ranking of transmitters that contribute most to harmful interference, supporting optimisation and spectrum planning.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21441
PDF
-
4DGS-Fixer: Generative Sparse-View 4D Gaussian Splatting with Iterative Refinement Guided by Video Diffusion Priors
2026-09-18 00:37:53Haitao Huang, Shenghao Zhao, Boyuan Tian, Shin-Fang Chng, Songlin Yang, Sheila Lim, Huangying Zhan, Yi Xu, Anyi Rao, Frank GuanAbstract
This paper addresses the challenges of dynamic scene synthesis from sparse-view videos. Existing methods employ geometric priors, adaptive optimization, or density-control strategies to improve 4D Gaussian modeling under sparse observations. However, they cannot fundamentally resolve the ill-posed problem caused by insufficient observations and missing scene information. Moreover, sparse-view 4D Gaussian Splatting (4DGS) often suffers from poor geometric initialization: with only a few input views, COLMAP typically reconstructs sparse and incomplete point clouds, leaving large scene regions without sufficient Gaussian support and making them difficult to recover through subsequent optimization. To address these limitations, we propose a novel iterative refinement framework based on a video diffusion model to improve the completeness and consistency of dynamic 4D scenes. Specifically, we first estimate multi-view depth maps and fuse them into dense point clouds to provide more complete geometric initialization for a dynamic 4DGS representation. We then employ a pretrained video restoration model to refine sequences rendered along novel camera trajectories at different time steps. The restored sequences serve as pseudo-supervision to regularize and iteratively refine the 4DGS representation. Experiments on a widely used benchmark dataset demonstrate that our method substantially outperforms existing baselines, achieving nearly a 2 dB PSNR improvement over the previous best-performing method.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21176
PDF
-
EnSol: an environment-aware graph neural network for molecular solubility prediction
2026-09-17 23:35:33Thao Nguyen, Saman Shafaei, Zhengyi Zhang, Huimin Zhao, Heng JiAbstract
Molecular solubility directly affects key aspects of molecular development such as reaction feasibility, formulation performance, separation efficiency, and solvent selection. However, experimental measurement across solutes, solvents, and temperatures remains costly and sparsely sampled. Existing computational models often rely on fixed-solvent assumptions, deterministic formulations, or simplified representations of solute-solvent interactions, limiting their ability to capture complex molecular interactions, continuous temperature effects, and experimental uncertainty. Here, we introduce EnSol, an environment-aware probabilistic framework for molecular solubility prediction. EnSol represents the solute and solvent as molecular graphs and learns separate representations for each before bringing them together through cross-attention to capture solute-solvent interactions. Temperature is incorporated directly into the solvent environment through feature-wise modulation, and a mixture density network predicts full solubility distributions to capture both temperature-dependent behavior and experimental uncertainty. On the independent SolProp and Leeds benchmark datasets, EnSol achieved Spearman correlations of 0.876 and 0.601, respectively, outperforming state-of-the-art solubility prediction models across both benchmarks. Beyond computational benchmarking, experimental validation across chemically diverse solute-solvent pairs showed that EnSol maintained strong predictive performance and supported reliable solvent ranking, achieving a Spearman correlation of 0.715. These results show that EnSol can support reliable solubility prediction and solvent selection across diverse chemical systems while accounting for predictive uncertainty.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21151
PDF
-
MarsRecon: Self-Supervised and Multimodal Surface Representations for Mars
2026-09-17 21:45:50Akshay Naik, Marius F. R. Juston, Jay MahajanAbstract
High-resolution orbital imagery offers a rich record of the Martian surface, but sparse geological labels limit supervised representation learning. We present MarsRecon, a geospatially aware pipeline for learning visual and multimodal representations from HiRISE observations of Olympus Mons. The pipeline calibrates NASA Planetary Data System products, extracts valid georeferenced patches, and trains a masked autoencoder on unlabeled imagery. Increasing input resolution and filtering invalid tokens reduced held-out reconstruction loss from 0.1751 to 0.1342 in the principal Stage A model series. We then freeze the visual encoder and align its features with observation text, coordinates, and local--global image context. The strongest current local-primary model achieves image-to-text recall@10 of 0.3787, text-to-image recall@10 of 0.9161, and local-to-global recall@10 of 0.4350 on the held-out test split. These results establish a working Mars-specific pretraining and retrieval pipeline; further crop-overlap controls and downstream geological evaluations are needed to assess the broader utility of its embeddings.
Abstract (translated)
URL
https://arxiv.org/abs/2609.22379
PDF
-
MAGIC: Marginal-Guided Compression with Optimal Transport for Efficient Visual Document Retrieval
2026-09-17 19:14:06Xu Yuan, Hua Liu, Wenqi Fan, Qing LiAbstract
Recent visual document retrieval (VDR) systems such as ColPali use multi-vector page embeddings, in which patch-level vectors enable fine-grained evidence matching but incur substantial index storage and MaxSim scoring overhead. Post-hoc merging offers a practical route to efficient VDR by reducing this cost without retraining the retriever, but its uniform reconstruction objectives are poorly aligned with the sparse, non-uniform patch usage induced by late-interaction retrieval. Under aggressive compression, this misalignment can preserve rarely used patches while concentrating retrieval activity on too few retained representatives. To address this misalignment, we propose Marginal-Guided Compression with Optimal Transport (MAGIC), a training-free post-hoc compressor for efficient retrieval with frozen multi-vector embeddings. MAGIC derives a MaxSim-induced compression surrogate and optimizes it through a two-marginal entropic optimal-transport formulation, where a retrieval-demand source marginal prioritizes high-use patches and a balanced target marginal regularizes retained-facet usage. Across ViDoRe benchmarks, keep ratios, and retrieval backbones, MAGIC consistently outperforms strong post-hoc compressors, with particularly large gains in the aggressive-compression regime; component ablations verify the complementary effects of its two marginals. We release the code at: this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21018
PDF
-
RBS-Attention: Radius-Bounded Sparse Prefill for Long-Context Large Language Models
2026-09-17 18:28:29Chuxu Song, Jiuqi Wei, Zhencan PengAbstract
Long-context large language model inference is increasingly limited by prefill, where dense self-attention processes the entire prompt before generation begins. Sparse block selection can reduce this cost, but a block centroid may hide a highly relevant token among many irrelevant ones. We call this failure mode mean dilution and propose RBS-Attention, a training-free sparse-prefill method with two complementary selection branches. A centroid base branch captures average relevance, while a rescue branch uses the maximum key-block radius and its prompt-, layer-, and head-dependent distribution to identify blocks at risk of underestimation. Independently thresholding the two branches and combining their masks controls the contribution of rescue blocks while preserving regular block-sparse FlashAttention execution. On H100 GPUs, RBS-Attention achieves 20.65$\times$ standalone prefill-attention speedup, 11.92$\times$ vLLM prefill-attention speedup, and 5.97$\times$ end-to-end time-to-first-token speedup at 128K on Qwen3-30B-A3B-Instruct-2507-FP8. On the dense Qwen3-32B model, it obtains 88.65 overall RULER accuracy versus 89.52 for dense attention; LongBench-v2, InfiniteBench, and Video-MME provide additional quality evaluation. Supporting experiments measure actual retention, compare selectors at matched density, and characterize block-size, threshold, and memory behavior. Together, these results support radius-adaptive dual-branch selection as an effective approach to long-context prefill.
Abstract (translated)
URL
https://arxiv.org/abs/2609.20971
PDF
-
FAMOS: Feed-Forward 3D Articulation Modeling from Sparse Observations
2026-09-17 17:59:40Kevin Qu, Tao Sun, Massimiliano Viola, Liyuan Zhu, Zhizhuo Zhou, Sayan Deb Sarkar, Konrad Schindler, Iro ArmeniAbstract
Modeling articulated objects from sparse monocular views is challenging because each observation reveals only partial geometry and motion evidence. Most feed-forward methods infer articulation from a single observation and therefore rely heavily on learned category-level shape priors. We present FAMOS, a feed-forward model that predicts movable-part segmentation and joint parameters from a sparse, unordered set of partial point clouds. Our model jointly reasons over multiple observations and naturally supports a variable number of inputs, including a single view. To aggregate articulation cues across observations, we introduce a Multi-state Articulation Transformer with alternating state-wise and global attention. We further propose an observed articulation span objective that supervises the motion range each part exhibits across the input observations, encouraging the model to leverage the full observation set. To overcome the limited scale and diversity of existing datasets, we introduce a procedural data generator that synthesizes self-annotated assets during training. Experiments on PartNet-Mobility, ACD, and ArtiCraft-10K demonstrate consistent improvements over both feed-forward and optimization-based baselines. Project page: this https URL
Abstract (translated)
URL
https://arxiv.org/abs/2609.20817
PDF
-
FunArt: Decoding Functional Structure and Articulation from Generative 3D Latents
2026-09-17 16:48:05Dennis Rotondi, Abdelrhman Werby, Kai O. ArrasAbstract
To operate effectively in human environments, robots must identify articulated objects, segment their movable and interactive parts, and estimate their kinematic models. Existing articulated scene representations typically recover kinematics from observed interactions, while methods operating on static scans often decouple articulation from functional interactive elements. We present FunArt, a framework that constructs articulation-aware functional 3D scene graphs from posed RGB-D observations captured in a single static configuration. FunArt reconstructs object instances, converts their fused geometry directly into the O-Voxel representation of TRELLIS.2, and exploits its frozen, sparse-compression VAE as a structural prior. A lightweight query-based decoder combines compact object-level latents with dense, surface-aligned features to jointly segment movable parts and functional interactive elements while estimating motion type, axis, origin, and range. On the Articulate3D dataset, FunArt achieves state-of-the-art performance across movable-part segmentation, articulation estimation, and functional-element segmentation, both with and without ground-truth object input. In the end-to-end setting, it outperforms the strongest baselines by 1.5 AP_{50} points for movable parts, 2.8 AP_{50} points under joint origin-and-axis constraints, and 6.7 AP_{50} points for functional elements. These results demonstrate that generative 3D latents encode actionable structural cues that can initialize robotic perception and planning before physical interaction.
Abstract (translated)
URL
https://arxiv.org/abs/2609.20673
PDF
