Searcharxiv⌕ Search

arXiv subjects

Jing Xiao

Publications and source records attributed to Jing Xiao.

At least 19 recordsLinked to original sources
arXiv2026

DGT-Map: Directional Global Traversability Mapping Utilizing Multi-Task Learning for Heterogeneous Vehicles

Off-road traversability is direction-dependent and vehicle specific, yet most global maps assign a single isotropic cost to each location. Existing learned estimators are also commonly trained independently for each vehicle; this preserves vehicle-specific behavior but prevents vehicles from sharing common terrain representations. DGT-MAP addresses both limitations through a self-supervised framework that learns global, directional, and vehicle-conditioned traversability costmaps from RGB-D observations and locomotion signals. A shared multi-task backbone learns common terrain features across training vehicles while vehicle-specific prediction heads preserve platform-dependent responses. At inference, DGT-MAP produces a heading-indexed costmap that can be used by a direction-aware planner. We evaluate DGT-MAP in simulation by integrating it into a Hybrid A* navigation stack and measuring downstream task success on challenging terrains, including slopes that are traversable downhill but not uphill and a ridge obstacle that is traversable by some vehicles, but not by others. Across evaluated tasks, DGT-MAP achieves the highest or tied-highest navigation success rate when compared against geometric, binary, and learned direction-agnostic baselines.

cs.RO↗
arXiv2026

Validation of Human Motor Development Scales with Humanoid Robots

We investigate how humanoid robots can learn a series of motor skills defined by human infant motor development scales, and we evaluate which scales are comprehensive enough to serve as self-contained guides for developing gross motor skills up to standing and walking. We represent each motor development scale as a developmental graph that guides the training of a humanoid robot. We then apply this graph within a reinforcement learning framework so that the robot learns skills following the natural phases of human motor development described by the scale. Our work demonstrates that graphs derived from three out of four popular pediatric motor scales can successfully guide humanoid robots to develop these skills. The fourth scale misses some intermediate milestones that the other scales have, which causes some failures. After adding a missing milestone in its graph, it also enables the robot to develop those skills successfully.

cs.RO↗
arXiv2026

PIVOT: Perplexity-Informed KD-to-RL Transition Scheduling for Vertical-Domain Few-Shot Distillation

Vertical-domain few-shot classification remains challenging for small language models, as limited supervision makes it difficult to acquire domain-specific decision knowledge. On-Policy Distillation (OPD) can improve teacher-guided adaptation by supervising student-generated rollouts, while GRPO-based reinforcement learning can further refine downstream predictions. However, existing KD-to-RL pipelines typically rely on globally fixed transition schedules, ignoring that different samples may require different amounts of teacher-guided acquisition before reward-driven refinement. We propose PIVOT (Perplexity-Informed Transition Optimization), a dynamic transition framework that routes samples between OPD and GRPO according to teacher-evaluated sequence perplexity. PIVOT moves low-perplexity samples to GRPO for reward-driven refinement while keeping high-perplexity samples under OPD for continued domain knowledge acquisition. Experiments on Banking77 and HWU64 show that PIVOT consistently outperforms continued OPD and globally synchronized OPD$\rightarrow$GRPO baselines under the same number of post-warm-up student optimization steps, achieving stronger downstream performance and more stable training dynamics.

cs.LG↗
arXiv2026

Asymmetric Scout-Worker Reconnaissance for Route Validation in Unknown Environments

This paper studies asymmetric scout-worker reconnaissance in unknown environments, where a small, agile autonomous scout explores routes for a larger worker robot that must visit an ordered sequence of goal locations. Because the scout has a smaller footprint and greater mobility, a scout-traversable route may be infeasible for the worker; worker feasibility must therefore be inferred from scout observations. This setting is not explicitly addressed by existing exploration and replanning methods, which typically assume a single traversability model and seek optimal paths for the same robot performing the exploration. We introduce a symbiotic scout-based framework that exploits the scout's superior mobility to explore only the portions of the unknown environment needed to identify worker-feasible path segments connecting the ordered goals. Evaluations in simulated and real-world settings demonstrate that the proposed approach validates feasible routes, repairs blocked segments with validated worker-feasible detours, and substantially reduces scout travel compared to baseline exploration and planning methods. A real-world indoor deployment further demonstrates the scout navigating narrow corridors to identify a worker-feasible route.

cs.RO↗
arXiv2026

IronLLM: Forging Compact Edge-Native Language Models for Real-Time Embodied Intelligence

We present IronLLM-0.6B, a 654M-parameter language model designed for efficient on-device inference. IronLLM-0.6B combines a hybrid attention architecture with X-MTP, a lightweight shared-KV multi-token prediction design that eliminates per-depth KV-cache replay and employs a lightweight verification head for rollback-free drafting, achieving a 1.48x decoding speedup. The model is pretrained on approximately 6.2 trillion tokens using a quality-oriented data pipeline and is further post-trained with Multi-Domain On-Policy Distillation to integrate capabilities from domain-specialized teachers. To better meet the low-latency requirements of on-device scenarios, IronLLM-0.6B adopts an Instruct-Only design. Evaluations show that IronLLM-0.6B achieves competitive performance relative to larger models such as Qwen3.5-0.8B and MiniCPM5-1B, while producing more concise responses on many tasks. We further present IronLLM-0.6B-Light, which replaces RMSNorm with Dynamic Tanh and simplifies several computationally expensive components to improve inference and quantization efficiency. Together, the IronLLM models provide an effective performance-efficiency trade-off for resource-constrained deployment.

cs.AI↗
arXiv2026

Multi-Modal Non-Prehensile Estimation of Physical Parameters via Press-and-Pull Tipping

Recovering physical properties of unknown objects through non-prehensile interaction is challenging because no single manipulation primitive reveals all relevant parameters. Planar pushing couples mass and friction, while conventional tipping cannot recover friction and may fail entirely when low-friction or curved-base objects slide or rotate instead of tipping. We introduce a multi-modal estimation framework that combines a sliding interaction with a press-and-pull tipping primitive to recover object mass, center of mass height, and surface friction. The press-and-pull interaction increases the object-table sliding threshold and stabilizes the pivot, enabling controlled tipping without any prior geometric object model. Wrist force/torque sensing, RGB-D perception, and robot proprioception are fused to estimate the physical parameters from the two complementary interaction modes. Experiments on an ABB IRB120 across four objects with varied geometry, mass, center of mass, and friction achieve low relative error, while successfully operating on curved-base objects that fail under conventional forward tipping. The results demonstrate that complementary non-prehensile interactions can recover a compact set of physical parameters without grasping, a prior object model, or learned interaction dynamics.

cs.RO↗
arXiv2026

Gripper-Aware Automatic Dense Packing of Irregular Objects

Automatic dense packing is widely desired in warehouse operations but remains a fundamental challenge in robotic manipulation. Existing work on irregular-object packing largely targets simulation with idealized contact, treating the object as an isolated rigid body. The gripper often enters as a discrete, post-hoc feasibility check, if considered at all, and the perception and contact drift accumulated during execution are not addressed. We present a closed-loop pipeline that integrates perception, gripper-aware placement optimization, and force-guided execution on a real manipulator. The optimizer represents the object together with the gripper as a single composite body of hierarchical sphere trees. It searches over five degrees of freedom on a GPU within a CMA-ES framework, with the vertical coordinate grounded analytically against the current heightmap. During execution, a force-monitored vertical descent stops on first contact. A post-release consolidation push then closes the residual lateral clearance that gripper-aware planning leaves behind. The container is re-perceived between placements so that drift does not accumulate. We validate the system on a Franka Emika Panda robot packing a 3D-printed set of flat, curved, and concave objects, and a YCB object subset. An ablation study isolates the contribution of gripper-aware optimization, the consolidation push, and mesh-derived geometry to end-to-end success, achieved density, and computational cost. We further benchmark against the heightmap-minimization method as a baseline representative of prior irregular-object packing work.

cs.RO↗
arXiv2026

Before the Tipping Point: Force-Guided Active Perception for Shape-Agnostic Estimation of 3D Centers of Mass

Estimating the 3D center of mass of unknown objects is challenging when grasping is infeasible, geometry is irregular, or mass distribution is uneven. We present a force-based method that estimates CoM height and mass from a single sub-critical tipping experiment by a robot manipulator. The robot applies a quasistatic elevated push and retract motion, using force-angle measurements recorded during tipping to identify parameters from the object trajectory. Our proposed push-retract cycle mitigates frictional bias, enabling generalized fitting. We experimentally validate our method using a robot manipulator with a six-axis force torque sensor on varying types of objects without prior shape information and without specific models. We also propose a method to prevent toppling, keeping the object in a sub-critical tipping regime by leveraging a safety margin. In experimental studies, our method recovers mass, CoM height, and toppling angle with relative errors below 5.0 percent across all unknown objects. This work demonstrates reliable 3D inertial parameter estimation under proper safety thresholds in tipping. Our proposed method informs and enables reliable non-prehensile manipulation and robotic grasping of challenging objects that were previously infeasible.

cs.RO↗
arXiv2026

Gradient-Update Mismatch: Rethinking Conflict-Free Training of Physics-Informed Neural Networks

Training Physics-Informed Neural Networks (PINNs) requires jointly optimizing physics residual and initial/boundary condition loss terms, which often induce conflicting gradients. Gradient surgery methods mitigate this issue by constructing directions from loss-specific gradients to reduce conflict before optimizer transformation. However, even when the constructed direction is conflict-free, this property may not be preserved after optimizer transformation. Let $a_t$ denote the direction constructed by gradient surgery, $u_t$ the optimizer proposal, and $\mathcal{C}_t$ the conflict-free cone induced by the loss-specific gradients. We show that modern optimizers can transform $a_t$ through mechanisms such as historical state, adaptive scaling, preconditioning, or decoupled weight decay, so $a_t \in \mathcal{C}_t$ does not generally imply $u_t \in \mathcal{C}_t$. We refer to this optimizer-induced discrepancy in conflict-freeness between $a_t$ and $u_t$ as Gradient-Update Mismatch (GUM). Accordingly, we propose Gradient-Update Alignment (GUA), which projects $u_t$ onto $\mathcal{C}_t$ to obtain the aligned update $p_t$ and applies $p_t$ to the parameters. When the optimizer maintains internal state, GUA further adjusts this state toward targets reconstructed from the applied update. We conduct extensive experiments and find that GUM is widespread across momentum, adaptive, and curvature-based optimizers, with conflict rates reaching up to 86.3%. Across all PINN settings, GUA achieves conflict-free applied updates and consistently improves various gradient surgery methods, reducing the relative $L_2$ error by up to 98.2% in individual settings. Data and code are available at https://github.com/JingXiao10/GUA.

cs.LG↗
arXiv2026

Motion-Aware Vision-Reference Alignment for Referring Multi-Object Tracking

Referring Multi-Object Tracking (RMOT) extends conventional multi-object tracking (MOT) by introducing natural language references for multi-modal fusion tracking. RMOT benchmarks only describe the object's appearance, relative positions, and initial motion states. This so-called static regulation fails to capture dynamic changes of the object motion, including velocity changes and motion direction shifts. This limitation not only causes a temporal discrepancy between static references and dynamic vision modality but also constrains multi-modal tracking performance. To address this limitation, we propose a novel motion-aware vision-reference alignment framework, named VMRMOT. VMRMOT introduces a motion modality derived from object dynamics to facilitate the alignment between the vision modality and language references. Specifically, we introduce motion-aware descriptions derived from object dynamic behaviors and encode them into motion features as the motion modality through multi-modal large language models (MLLMs). We further design a Vision-Motion-Reference Alignment (VMRA) module to hierarchically align visual queries with motion and reference cues, enhancing their cross-modal consistency. In addition, a Motion-Guided Prediction Head (MGPH) is developed to explore motion modality to enhance the performance of the prediction head. To the best of our knowledge, VMRMOT is the first motion-aware vision-reference alignment framework for the RMOT task. Extensive experiments on multiple RMOT benchmarks demonstrate that VMRMOT outperforms existing state-of-the-art methods. The code is available at https://github.com/Kroery/VMRMOT.

cs.CV↗
arXiv2026

Observe Less, Understand More: Cost-aware Cross-scale Observation for Remote Sensing Understanding

Remote sensing understanding inherently requires multi-resolution observation, since different targets and application tasks demand different levels of spatial detail. While low-resolution (LR) imagery enables efficient global observation, high-resolution (HR) imagery provides critical local details at a much higher acquisition cost and with limited coverage. This motivates a cross-scale sensing strategy that selectively acquires HR imagery guided by LR-based global perception to improve task performance under constrained cost. Existing HR sampling methods typically make selection decisions from isolated LR patches, thereby ignoring fine-grained intra-patch importance and cross-patch contextual interactions, leading to fragmented feature representation and suboptimal scene reasoning under sparse HR observations. To address this issue, we formulate cross-scale remote sensing understanding as a unified cost-aware problem that couples fine-grained HR sampling with cross-patch representation prediction, enabling more effective task reasoning with fewer HR observations. Furthermore, we present GL-10M, a high- and low-resolution dataset with nearly 100,000 scene pairs and 10 million images for large-scale cross-resolution pretraining. Extensive experiments on recognition and retrieval tasks show that our method consistently achieves a superior performance-cost trade-off. The code is publicly available at https://github.com/xzhacc/CrossSO.

cs.CV↗
arXiv2026

SA-GEM: Scale-Adaptive and Geospatial Evidence-Modulated Token Pruning for Efficient Remote Sensing Large Vision-Language Models

RS-LVLMs have advanced multimodal understanding of Earth observation imagery, yet their performance is fundamentally constrained by high-resolution processing, as visual token counts grow quadratically with linear input resolution while important visual evidence is inherently sparse and increasingly diluted across the expanded sequence. Existing token pruning methods largely rely on scale-agnostic resolution policies and isolated importance cues, limiting task-aligned granularity adaptation and holistic evidence preservation. To address this, we present Scale-Adaptive and Geospatial Evidence-Modulated Token Pruning (SA-GEM), a plug-and-play framework that unifies task-adaptive token granularity allocation with holistic geospatial token importance modulation. Specifically, a lightweight router selects the resolution based on query-dependent token granularity, while a token importance modulator jointly models task relevance, spatial structure, and local redundancy to preserve holistic geospatial evidence. We show that higher resolution is not universally beneficial and, once sufficient granularity is reached, token quality matters more than token quantity. Experiments across various benchmarks demonstrate that SA-GEM achieves consistent gains in both accuracy and efficiency over existing pruning methods. On XLRS-Bench, it surpasses GeoLLaVA-8K by 2.3% in accuracy with a 2.4 times total inference speedup.

cs.CV↗
arXiv2026

LongEarth-R1: Benchmarking and Aligning Vision-Language Models for Long-Horizon Earth Observation Reasoning

Long-horizon Earth observation reasoning requires models to organize multi-stage geographic evolution, localize spatial changes, detect temporal anomalies, and infer future from extended image sequences. However, existing remote sensing vision-language models mainly focus on isolated images, image pairs, or short sequences, limiting reliable grounding in the relevant frames and regions. We introduce LongEarth-Bench, a benchmark containing approximately 120k question-answering samples derived from 117k unique images. Its sequences average 15.14 frames and extend to 30 frames, covering 12 tasks across evolution summarization, spatial reasoning, anomaly identification, and logical prediction. A 30k-sample subset further provides structured reasoning traces linking key frames and changed regions to final answers. We develop LongEarth through supervised fine-tuning with explicit sequence identifiers and structured chain-of-thought supervision. Building on LongEarth, LongEarth-R1 applies group relative policy optimization with format, temporal, and spatial rewards. LongEarth-R1 achieves the best results on all 12 long-sequence tasks while remaining competitive on standard remote sensing benchmarks.

cs.AI↗
arXiv2026

Decoupled Similarity for Task-Aware Token Pruning in Large Vision-Language Models

Token pruning has emerged as an effective approach to reduce the substantial computational overhead of Large Vision-Language Models (LVLMs) by discarding less informative visual tokens while preserving performance. However, existing methods typically rely on individual attention sources from different LVLM components, resulting in incomplete and suboptimal pruning decisions due to biased attention distributions. To address this problem, we propose DeSAP, a novel Decoupled Similarity-Aware Pruning method for precise, task-aware token pruning within the visual encoder. Specifically, DeSAP introduces a decoupled similarity to capture fine-grained cross-modal relevance between visual features and text tokens, providing explicit task-related guidance for pruning. By integrating decoupled similarity with visual saliency signals derived from visual attention, DeSAP performs token pruning under the guidance of both task-related and visual cues, enabling robust pruning even under aggressive pruning ratios. Extensive experiments across diverse benchmarks and architectures show that DeSAP consistently outperforms SOTA methods in both accuracy and efficiency. On LLaVA-1.5-7B, DeSAP achieves a 10 times FLOPs reduction and a 2.3 times prefill speedup by retaining only 11.1% of visual tokens, while maintaining 98.1% of the original performance.

cs.CV↗
arXiv2026

HydroFirn: A numerical model for large-scale multidimensional firn hydrology

Observations show the multidimensional dynamics of meltwater and distribution of ice layers in the firn on the Greenland Ice Sheet. However, state-of-the-art large-scale models for firn hydrology are essentially one-dimensional, limiting their ability to explain observed datasets and failing to reduce uncertainty in surface mass balance and sea-level rise estimates. Here, we present a large-scale, multidimensional, multiphase, and thermomechanical model to simulate firn hydrology. The model is highly efficient due to a novel algorithm in which an extra equation for pressure is solved only in saturated regions. Furthermore, the model can apply spatially heterogeneous boundary conditions to the unsaturated-saturated domain and allows for the dynamic formation of fully impermeable ice layers. The numerical results show excellent comparisons against analytic solutions to one- and two-dimensional problems that involve coupled unsaturated-saturated flows, thermodynamics, and phase change. We further apply the model to investigate field data from southwest Greenland and find that lateral heterogeneities strongly influence the depth of melt percolation and ice layer formation. Improved understanding of these local, multidimensional processes will provide physics-based constraints on firn densification, reduce uncertainty in converting altimetric elevation change to mass change, and improve estimates of freshwater fluxes to the ocean under a warming climate.

physics.geo-ph↗
arXiv2026

Developing Combined Manipulation and Locomotion Skills with Interaction Representation and Skill Composition

This paper addresses how to enable a humanoid robot to learn motion policies based on developmental principles and combine policies to create more sophisticated and useful behaviors. Specifically, we present an approach to (1) learning a whole-body reaching and grasping policy and (2) combining it and a standing-up and walking policy to compose a more complex policy of manipulation and locomotion: grasping, standing up, and walking. In (1), our method draws inspiration from harmonic analysis and adopts cubic harmonics as weights to represent the hand-object spatial relationship via spatial convolution. Utilizing an intra-episode finger joint decoupling curriculum based on developmental principles, a robot can autonomously learn a generalizable grasping policy without relying on external datasets or pretrained models. In (2), our method combines the grasping policy with a separately learned getting-up policy by providing both policies with their respective observation vectors and using hand-object interaction scores to determine when each policy should control which robot joints. Our results show a 93% zero-shot success rate for grasping unseen objects and a 96-100% success rate for standing up while holding the object. Our work also demonstrates that combining different policies is only effective if each policy learning happens on the same whole humanoid body even if a policy (such as for locomotion) does not seem to need all the body parts (such as fingers).

cs.RO↗
arXiv2026

Rethinking LLM-as-a-Judge: Representation-as-a-Judge with Small Language Models via Semantic Capacity Asymmetry

Large language models (LLMs) are widely used as reference-free evaluators via prompting, but this "LLM-as-a-Judge" paradigm is costly, opaque, and sensitive to prompt design. In this work, we investigate whether smaller models can serve as efficient evaluators by leveraging internal representations instead of surface generation. We uncover a consistent empirical pattern: small LMs, despite with weak generative ability, encode rich evaluative signals in their hidden states. This motivates us to propose the Semantic Capacity Asymmetry Hypothesis: evaluation requires significantly less semantic capacity than generation and can be grounded in intermediate representations, suggesting that evaluation does not necessarily need to rely on large-scale generative models but can instead leverage latent features from smaller ones. Our findings motivate a paradigm shift from LLM-as-a-Judge to Representation-as-a-Judge, a decoding-free evaluation strategy that probes internal model structure rather than relying on prompted output. We instantiate this paradigm through INSPECTOR, a probing-based framework that predicts aspect-level evaluation scores from small model representations. Experiments on reasoning benchmarks (GSM8K, MATH, GPQA) show that INSPECTOR substantially outperforms prompting-based small LMs and closely approximates full LLM judges, while offering a more efficient, reliable, and interpretable alternative for scalable evaluation. The code and data are available at: https://github.com/zhuochunli/Representation-as-a-judge

cs.CL↗
arXiv2026

Sentinel: Decoding Context Utilization via Attention Probing for Efficient LLM Context Compression

Retrieval-augmented generation (RAG) often suffers from long and noisy retrieved contexts. Existing context compression methods typically rely on heuristic relevance estimation or supervised compression models rather than on how LLMs utilize retrieved context during inference. We propose Sentinel, a lightweight sentence-level compression framework that decodes inference-time contextual utilization behaviors from head-wise attention patterns of frozen LLMs. To ground supervision in retrieval-dependent answering behavior, Sentinel trains a lightweight probe using QA examples where the model succeeds only when retrieved context is available. Sentinel performs compression using only a single non-autoregressive forward pass without dedicated compression training or autoregressive scoring. Empirically, we find that effective contextual utilization signals remain accessible even in compact proxy models. On LongBench, Sentinel with a 0.5B proxy model achieves up to 5$\times$ compression while attaining question-answering performance competitive with compression methods built on 7B-scale models. Despite being trained only on English QA data, Sentinel also generalizes effectively to Chinese and out-of-domain settings.

cs.CL↗