arXiv subjects

Xinlei Chen

Publications and source records attributed to Xinlei Chen.

At least 55 records · Page 3Linked to original sources

arXiv2026

MARSHAL: Incentivizing Multi-Agent Reasoning via Self-Play with Strategic LLMs

Developing Large Language Models (LLMs) to cooperate and compete effectively within multi-agent systems (MASs) is a critical step towards more advanced intelligence. While reinforcement learning (RL) has proven effective for enhancing reasoning in single-agent tasks, its extension to multi-turn, multi-agent scenarios remains underexplored due to the challenges of long-horizon credit assignment and agent-specific advantage estimation. To address these challenges, we introduce MARSHAL, an end-to-end RL framework that incentivizes Multi-Agent Reasoning through Self-play witH strAtegic LLMs in both cooperative and competitive games. MARSHAL features a turn-level advantage estimator that aligns learning signals with each interaction for credit assignment, and an agent-specific advantage normalization to stabilize multi-agent training. By learning with self-play across cooperative and competitive games, MARSHAL agents trained from Qwen3-4B develop strong strategic abilities, with up to 28.7% performance improvements in held-out games. More importantly, the capability acquired through self-play generalizes beyond games, yielding consistent performance gains of MASs in reasoning benchmarks. When integrated into leading MASs, our MARSHAL agent achieves significant zero-shot performance gains of up to 10.0% on AIME, 7.6% on GPQA-Diamond, and 3.5% on average across all benchmarks. These results establish self-play in strategic games as a powerful approach for developing generalizable multi-agent reasoning capabilities in LLMs.

cs.AI↗

arXiv2026

What Can RL Bring to VLA Generalization? An Empirical Study

Large Vision-Language Action (VLA) models have shown significant potential for embodied AI. However, their predominant training via supervised fine-tuning (SFT) limits generalization due to susceptibility to compounding errors under distribution shifts. Reinforcement learning (RL) offers a path to overcome these limitations by optimizing for task objectives via trial-and-error, yet a systematic understanding of its specific generalization benefits for VLAs compared to SFT is lacking. To address this, our study introduces a comprehensive benchmark for evaluating VLA generalization and systematically investigates the impact of RL fine-tuning across diverse visual, semantic, and execution dimensions. Our extensive experiments reveal that RL fine-tuning, particularly with PPO, significantly enhances generalization in semantic understanding and execution robustness over SFT, while maintaining comparable visual robustness. We identify PPO as a more effective RL algorithm for VLAs than LLM-derived methods like DPO and GRPO. We also develop a simple recipe for efficient PPO training on VLAs, and demonstrate its practical utility for improving VLA generalization. The project page is at https://rlvla.github.io

cs.LG↗

arXiv2026

JuggleRL: Mastering Ball Juggling with a Quadrotor via Deep Reinforcement Learning

Aerial robots interacting with objects must perform precise, contact-rich maneuvers under uncertainty. In this paper, we study the problem of aerial ball juggling using a quadrotor equipped with a racket, a task that demands accurate timing, stable control, and continuous adaptation. We propose JuggleRL, the first reinforcement learning-based system for aerial juggling. It learns closed-loop policies in large-scale simulation using systematic calibration of quadrotor and ball dynamics to reduce the sim-to-real gap. The training incorporates reward shaping to encourage racket-centered hits and sustained juggling, as well as domain randomization over ball position and coefficient of restitution to enhance robustness and transferability. The learned policy outputs mid-level commands executed by a low-level controller and is deployed zero-shot on real hardware, where an enhanced perception module with a lightweight communication protocol reduces delays in high-frequency state estimation and ensures real-time control. Experiments show that JuggleRL achieves an average of $311$ hits over $10$ consecutive trials in the real world, with a maximum of $462$ hits observed, far exceeding a model-based baseline that reaches at most $14$ hits with an average of $3.1$. Moreover, the policy generalizes to unseen conditions, successfully juggling a lighter $5$ g ball with an average of $145.9$ hits. This work demonstrates that reinforcement learning can empower aerial robots with robust and stable control in dynamic interaction tasks.

cs.RO↗

arXiv2026

Aerial World Model for Long-horizon Visual Generation and Navigation in 3D Space

Unmanned aerial vehicles (UAVs) have emerged as powerful embodied agents. One of the core abilities is autonomous navigation in large-scale three-dimensional environments. Existing navigation policies, however, are typically optimized for low-level objectives such as obstacle avoidance and trajectory smoothness, lacking the ability to incorporate high-level semantics into planning. To bridge this gap, we propose ANWM, an aerial navigation world model that predicts future visual observations conditioned on past frames and actions, thereby enabling agents to rank candidate trajectories by their semantic plausibility and navigational utility. ANWM is trained on 4-DoF UAV trajectories and introduces a physics-inspired module: Future Frame Projection (FFP), which projects past frames into future viewpoints to provide coarse geometric priors. This module mitigates representational uncertainty in long-distance visual generation and captures the mapping between 3D trajectories and egocentric observations. Empirical results demonstrate that ANWM significantly outperforms existing world models in long-distance visual forecasting and improves UAV navigation success rates in large-scale environments.

cs.RO↗

arXiv2025

PhysFire-WM: A Physics-Informed World Model for Emulating Fire Spread Dynamics

Fine-grained fire prediction plays a crucial role in emergency response. Infrared images and fire masks provide complementary thermal and boundary information, yet current methods are predominantly limited to binary mask modeling with inherent signal sparsity, failing to capture the complex dynamics of fire. While world models show promise in video generation, their physical inconsistencies pose significant challenges for fire forecasting. This paper introduces PhysFire-WM, a Physics-informed World Model for emulating Fire spread dynamics. Our approach internalizes combustion dynamics by encoding structured priors from a Physical Simulator to rectify physical discrepancies, coupled with a Cross-task Collaborative Training strategy (CC-Train) that alleviates the issue of limited information in mask-based modeling. Through parameter sharing and gradient coordination, CC-Train effectively integrates thermal radiation dynamics and spatial boundary delineation, enhancing both physical realism and geometric accuracy. Extensive experiments on a fine-grained multimodal fire dataset demonstrate the superior accuracy of PhysFire-WM in fire spread prediction. Validation underscores the importance of physical priors and cross-task collaboration, providing new insights for applying physics-informed world models to disaster prediction.

cs.CV↗

arXiv2025

VAEER: Visual Attention-Inspired Emotion Elicitation Reasoning

Images shared online strongly influence emotions and public well-being. Understanding the emotions an image elicits is therefore vital for fostering healthier and more sustainable digital communities, especially during public crises. We study Visual Emotion Elicitation (VEE), predicting the set of emotions that an image evokes in viewers. We introduce VAEER, an interpretable multi-label VEE framework that combines attention-inspired cue extraction with knowledge-grounded reasoning. VAEER isolates salient visual foci and contextual signals, aligns them with structured affective knowledge, and performs per-emotion inference to yield transparent, emotion-specific rationales. Across three heterogeneous benchmarks, including social imagery and disaster-related photos, VAEER achieves state-of-the-art results with up to 19% per-emotion improvements and a 12.3% average gain over strong CNN and VLM baselines. Our findings highlight interpretable multi-label emotion elicitation as a scalable foundation for responsible visual media analysis and emotionally sustainable online ecosystems.

cs.CV↗

arXiv2025

QUIDS: Quality-informed Incentive-driven Multi-agent Dispatching System for Mobile Crowdsensing

This paper addresses the challenge of achieving optimal Quality of Information (QoI) in non-dedicated vehicular mobile crowdsensing (NVMCS) systems. The key obstacles are the interrelated issues of sensing coverage, sensing reliability, and the dynamic participation of vehicles. To tackle these, we propose QUIDS, a QUality-informed Incentive-driven multi-agent Dispatching System, which ensures high sensing coverage and reliability under budget constraints. QUIDS introduces a novel metric, Aggregated Sensing Quality (ASQ), to quantitatively capture QoI by integrating both coverage and reliability. We also develop a Mutually Assisted Belief-aware Vehicle Dispatching algorithm that estimates sensing reliability and allocates incentives under uncertainty, further improving ASQ. Evaluation using real-world data from a metropolitan NVMCS deployment shows QUIDS improves ASQ by 38% over non-dispatching scenarios and by 10% over state-of-the-art methods. It also reduces reconstruction map errors by 39-74% across algorithms. By jointly optimizing coverage and reliability via a quality-informed incentive mechanism, QUIDS enables low-cost, high-quality urban monitoring without dedicated infrastructure, applicable to smart-city scenarios like traffic and environmental sensing.

cs.CV↗

arXiv2025

Event Camera Meets Mobile Embodied Perception: Abstraction, Algorithm, Acceleration, Application

With the increasing complexity of mobile device applications, these devices are evolving toward high agility. This shift imposes new demands on mobile sensing, particularly in achieving high-accuracy and low-latency. Event-based vision has emerged as a disruptive paradigm, offering high temporal resolution and low latency, making it well-suited for high-accuracy and low-latency sensing tasks on high-agility platforms. However, the presence of substantial noisy events, lack of stable, persistent semantic information, and large data volume pose challenges for event-based data processing on resource-constrained mobile devices. This paper surveys the literature from 2014 to 2025 and presents a comprehensive overview of event-based mobile sensing, encompassing its fundamental principles, event \textit{abstraction} methods, \textit{algorithm} advancements, and both hardware and software \textit{acceleration} strategies. We discuss key \textit{applications} of event cameras in mobile sensing, including visual odometry, object tracking, optical flow, and 3D reconstruction, while highlighting challenges associated with event data processing, sensor fusion, and real-time deployment. Furthermore, we outline future research directions, such as improving the event camera with advanced optics, leveraging neuromorphic computing for efficient processing, and integrating bio-inspired algorithms. To support ongoing research, we provide an open-source \textit{Online Sheet} with recent developments. We hope this survey serves as a reference, facilitating the adoption of event-based vision across diverse applications.

cs.RO↗

arXiv2025

STeP-Diff: Spatio-Temporal Physics-Informed Diffusion Models for Mobile Fine-Grained Pollution Forecasting

Fine-grained air pollution forecasting is crucial for urban management and the development of healthy buildings. Deploying portable sensors on mobile platforms such as cars and buses offers a low-cost, easy-to-maintain, and wide-coverage data collection solution. However, due to the random and uncontrollable movement patterns of these non-dedicated mobile platforms, the resulting sensor data are often incomplete and temporally inconsistent. By exploring potential training patterns in the reverse process of diffusion models, we propose Spatio-Temporal Physics-Informed Diffusion Models (STeP-Diff). STeP-Diff leverages DeepONet to model the spatial sequence of measurements along with a PDE-informed diffusion model to forecast the spatio-temporal field from incomplete and time-varying data. Through a PDE-constrained regularization framework, the denoising process asymptotically converges to the convection-diffusion dynamics, ensuring that predictions are both grounded in real-world measurements and aligned with the fundamental physics governing pollution dispersion. To assess the performance of the system, we deployed 59 self-designed portable sensing devices in two cities, operating for 14 days to collect air pollution data. Compared to the second-best performing algorithm, our model achieved improvements of up to 89.12% in MAE, 82.30% in RMSE, and 25.00% in MAPE, with extensive evaluations demonstrating that STeP-Diff effectively captures the spatio-temporal dependencies in air pollution fields.

cs.LG↗

arXiv2025

FireSentry: A Multi-Modal Spatio-temporal Benchmark Dataset for Fine-Grained Wildfire Spread Forecasting

Fine-grained wildfire spread prediction is crucial for enhancing emergency response efficacy and decision-making precision. However, existing research predominantly focuses on coarse spatiotemporal scales and relies on low-resolution satellite data, capturing only macroscopic fire states while fundamentally constraining high-precision localized fire dynamics modeling capabilities. To bridge this gap, we present FireSentry, a provincial-scale multi-modal wildfire dataset characterized by sub-meter spatial and sub-second temporal resolution. Collected using synchronized UAV platforms, FireSentry provides visible and infrared video streams, in-situ environmental measurements, and manually validated fire masks. Building on FireSentry, we establish a comprehensive benchmark encompassing physics-based, data-driven, and generative models, revealing the limitations of existing mask-only approaches. Our analysis proposes FiReDiff, a novel dual-modality paradigm that first predicts future video sequences in the infrared modality, and then precisely segments fire masks in the mask modality based on the generated dynamics. FiReDiff achieves state-of-the-art performance, with video quality gains of 39.2% in PSNR, 36.1% in SSIM, 50.0% in LPIPS, 29.4% in FVD, and mask accuracy gains of 3.3% in AUPRC, 59.1% in F1 score, 42.9% in IoU, and 62.5% in MSE when applied to generative models. The FireSentry benchmark dataset and FiReDiff paradigm collectively advance fine-grained wildfire forecasting and dynamic disaster simulation. The processed benchmark dataset is publicly available at: https://github.com/Munan222/FireSentry-Benchmark-Dataset.

cs.CV↗

arXiv2025

Count Every Rotation and Every Rotation Counts: Exploring Drone Dynamics via Propeller Sensing

As drone-based applications proliferate, paramount contactless sensing of airborne drones from the ground becomes indispensable. This work demonstrates concentrating on propeller rotational speed will substantially improve drone sensing performance and proposes an event-camera-based solution, \sysname. \sysname features two components: \textit{Count Every Rotation} achieves accurate, real-time propeller speed estimation by mitigating ultra-high sensitivity of event cameras to environmental noise. \textit{Every Rotation Counts} leverages these speeds to infer both internal and external drone dynamics. Extensive evaluations in real-world drone delivery scenarios show that \sysname achieves a sensing latency of 3$ms$ and a rotational speed estimation error of merely 0.23\%. Additionally, \sysname infers drone flight commands with 96.5\% precision and improves drone tracking accuracy by over 22\% when combined with other sensing modalities. \textit{ Demo: {\color{blue}https://eventpro25.github.io/EventPro/.} }

cs.RO↗

arXiv2025

AirCopBench: A Benchmark for Multi-drone Collaborative Embodied Perception and Reasoning

Multimodal Large Language Models (MLLMs) have shown promise in single-agent vision tasks, yet benchmarks for evaluating multi-agent collaborative perception remain scarce. This gap is critical, as multi-drone systems provide enhanced coverage, robustness, and collaboration compared to single-sensor setups. Existing multi-image benchmarks mainly target basic perception tasks using high-quality single-agent images, thus failing to evaluate MLLMs in more complex, egocentric collaborative scenarios, especially under real-world degraded perception conditions.To address these challenges, we introduce AirCopBench, the first comprehensive benchmark designed to evaluate MLLMs in embodied aerial collaborative perception under challenging perceptual conditions. AirCopBench includes 14.6k+ questions derived from both simulator and real-world data, spanning four key task dimensions: Scene Understanding, Object Understanding, Perception Assessment, and Collaborative Decision, across 14 task types. We construct the benchmark using data from challenging degraded-perception scenarios with annotated collaborative events, generating large-scale questions through model-, rule-, and human-based methods under rigorous quality control. Evaluations on 40 MLLMs show significant performance gaps in collaborative perception tasks, with the best model trailing humans by 24.38% on average and exhibiting inconsistent results across tasks. Fine-tuning experiments further confirm the feasibility of sim-to-real transfer in aerial collaborative perception and reasoning.

cs.CV↗

arXiv2025

Progressive Supernet Training for Efficient Visual Autoregressive Modeling

Visual Auto-Regressive (VAR) models significantly reduce inference steps through the "next-scale" prediction paradigm. However, progressive multi-scale generation incurs substantial memory overhead due to cumulative KV caching, limiting practical deployment. We observe a scale-depth asymmetric dependency in VAR: early scales exhibit extreme sensitivity to network depth, while later scales remain robust to depth reduction. Inspired by this, we propose VARiant: by equidistant sampling, we select multiple subnets ranging from 16 to 2 layers from the original 30-layer VAR-d30 network. Early scales are processed by the full network, while later scales utilize subnet. Subnet and the full network share weights, enabling flexible depth adjustment within a single model. However, weight sharing between subnet and the entire network can lead to optimization conflicts. To address this, we propose a progressive training strategy that breaks through the Pareto frontier of generation quality for both subnets and the full network under fixed-ratio training, achieving joint optimality. Experiments on ImageNet demonstrate that, compared to the pretrained VAR-d30 (FID 1.95), VARiant-d16 and VARiant-d8 achieve nearly equivalent quality (FID 2.05/2.12) while reducing memory consumption by 40-65%. VARiant-d2 achieves 3.5 times speedup and 80% memory reduction at moderate quality cost (FID 2.97). In terms of deployment, VARiant's single-model architecture supports zero-cost runtime depth switching and provides flexible deployment options from high quality to extreme efficiency, catering to diverse application scenarios.

cs.CV↗

arXiv2025

SN 2024aecx: A double-peaked rapidly evolving Type IIb supernova at 11 Mpc

We present the results of low-resolution spectroscopic and densely sampled multi-band photometric follow-up of supernova (SN) 2024aecx. The SN was discovered in the spiral galaxy NGC 3521 (distance $\sim$11 Mpc) within a day after the explosion. The early spectra of SN 2024aecx show a weak signature of hydrogen lines, which disappeared in $\sim$30 days after the explosion. Light curves in all bands show a distinct feature of two peaks, and the first peak is likely due to the shock cooling emission. The early phase light curve evolution of SN 2024aecx has similarity with the typical Type IIb events, but the decay rate in different bands (e.g., $\rm Δm_{15}$ = 1.60 $\pm$ 0.05 mag, $g$-band) is significantly faster in the post-peak phase. It attained the secondary maximum in $\sim$19 days ($g$-band) with a peak absolute magnitude of M$_{g}$ = -17.94 $\pm$ 0.10 mag. SN 2024aecx colors trend redder in early epochs ($<$8 days), followed by a duration in which it grows bluer, then later gets redder again $>$20 days after explosion. The analytical model fitting to the light curves reveals an envelope mass and progenitor radii in the range of $\sim$0.03 - 0.24 $M_\odot$ and $\sim$169 - 200 $R_\odot$, respectively. Modeling of the pseudo-bolometric light curve suggests that synthesized $^{56}$Ni in the explosion was $\sim$0.15 M$_{\odot}$ with ejecta mass and kinetic energy of $\sim$0.7 M$_{\odot}$ and $\sim$0.16 $\times$ 10$^{51}$ erg, respectively. The observational properties and modeling indicate that the SN~2024aecx progenitor belongs to the extended progenitor category.

astro-ph.HE↗

arXiv2025

mmE-Loc: Facilitating Accurate Drone Landing with Ultra-High-Frequency Localization

For precise, efficient, and safe drone landings, ground platforms should real-time, accurately locate descending drones and guide them to designated spots. While mmWave sensing combined with cameras improves localization accuracy, lower sampling frequency of traditional frame cameras compared to mmWave radar creates bottlenecks in system throughput. In this work, we upgrade traditional frame camera with event camera, a novel sensor that harmonizes in sampling frequency with mmWave radar within ground platform setup, and introduce mmE-Loc, a high-precision, low-latency ground localization system designed for precise drone landings. To fully exploit the \textit{temporal consistency} and \textit{spatial complementarity} between these two modalities, we propose two innovative modules: \textit{(i)} the Consistency-instructed Collaborative Tracking module, which further leverages the drone's physical knowledge of periodic micro-motions and structure for accurate measurements extraction, and \textit{(ii)} the Graph-informed Adaptive Joint Optimization module, which integrates drone motion information for efficient sensor fusion and drone localization. Real-world experiments conducted in landing scenarios with a drone delivery company demonstrate that mmE-Loc significantly outperforms state-of-the-art methods in both accuracy and latency.

cs.RO↗

arXiv2025

Open3D-VQA: A Benchmark for Comprehensive Spatial Reasoning with Multimodal Large Language Model in Open Space

Spatial reasoning is a fundamental capability of multimodal large language models (MLLMs), yet their performance in open aerial environments remains underexplored. In this work, we present Open3D-VQA, a novel benchmark for evaluating MLLMs' ability to reason about complex spatial relationships from an aerial perspective. The benchmark comprises 73k QA pairs spanning 7 general spatial reasoning tasks, including multiple-choice, true/false, and short-answer formats, and supports both visual and point cloud modalities. The questions are automatically generated from spatial relations extracted from both real-world and simulated aerial scenes. Evaluation on 13 popular MLLMs reveals that: 1) Models are generally better at answering questions about relative spatial relations than absolute distances, 2) 3D LLMs fail to demonstrate significant advantages over 2D LLMs, and 3) Fine-tuning solely on the simulated dataset can significantly improve the model's spatial reasoning performance in real-world scenarios. We release our benchmark, data generation pipeline, and evaluation toolkit to support further research: https://github.com/EmbodiedCity/Open3D-VQA.code.

cs.CV↗

arXiv2025

SN 2024gy: Multi-epoch Spectroscopic Features Suggestive of Delayed Detonation in a Type Ia Supernova

We present photometric and spectroscopic observations of SN 2024gy, a Type Ia supernova (SN Ia) exhibiting high-velocity features (HVFs) in its early-time spectra. This SN reaches a peak $B$-band magnitude of $-19.25 \pm 0.29$ mag and subsequently declines by $Δm_{15}(B) \approx 1.12$ mag, consistent with the luminosity-width relation characteristic of normal SNe Ia. Based on the peak thermal luminosity of $(1.2 \pm 0.3) \times 10^{43}$ erg s$^{-1}$, we estimate that $0.57 \pm 0.14~\rm M_{\odot}$ of $^{56}$Ni was synthesized during the explosion. Our dense early spectral monitoring revealed significant velocity disparities within the ejecta. Notably, absorption features from the Ca II near-infrared triplet were observed at velocities exceeding 25,000 km s$^{-1}$, while the Si II $λ$6355 line velocity at the same epoch was significantly lower at $\sim$ 16,000 km s$^{-1}$. This velocity disparity likely reflects distinct ionization states of intermediate-mass elements in the outermost layers. The prominent Ca II HVFs may originate from ionization suppression within the highest-velocity ejecta, potentially indicative of minimal hydrogen mixing in a delayed-detonation explosion scenario. Additionally, the Ni/Fe ratio derived from the nebular spectrum of SN 2024gy provides further support for this model.

astro-ph.HE↗

arXiv2025

Balanced Token Pruning: Accelerating Vision Language Models Beyond Local Optimization

Large Vision-Language Models (LVLMs) have shown impressive performance across multi-modal tasks by encoding images into thousands of tokens. However, the large number of image tokens results in significant computational overhead, and the use of dynamic high-resolution inputs further increases this burden. Previous approaches have attempted to reduce the number of image tokens through token pruning, typically by selecting tokens based on attention scores or image token diversity. Through empirical studies, we observe that existing methods often overlook the joint impact of pruning on both the current layer's output (local) and the outputs of subsequent layers (global), leading to suboptimal pruning decisions. To address this challenge, we propose Balanced Token Pruning (BTP), a plug-and-play method for pruning vision tokens. Specifically, our method utilizes a small calibration set to divide the pruning process into multiple stages. In the early stages, our method emphasizes the impact of pruning on subsequent layers, whereas in the deeper stages, the focus shifts toward preserving the consistency of local outputs. Extensive experiments across various LVLMs demonstrate the broad effectiveness of our approach on multiple benchmarks. Our method achieves a 78% compression rate while preserving 96.7% of the original models' performance on average. Our code is available at https://github.com/EmbodiedCity/NeurIPS2025-Balanced-Token-Pruning.

cs.CV↗