arXiv subjects

Chaoran Zhang

Publications and source records attributed to Chaoran Zhang.

15 recordsLinked to original sources

arXiv2026

GSAR: Goal-State-Anchor Rewards for Mobile GUI Agents with Self-Evolving Data Synthesis

Vision-Language Models (VLMs) based GUI agents stand to benefit significantly from online reinforcement learning (RL). However, their training is bottlenecked by two fundamental issues: current data synthesis methods for GUI Agents rely on specific environments and struggle to generate diverse data, while existing evaluators either suffer from limited scalability or provide inaccurate and unreliable reward signals. To overcome these challenges, we introduce GSAR (Goal-State-Anchor Reward), a RL reward framework that supports scalable task generation and delivers reliable reward signals for stable and efficient policy optimization. Our approach features self-evolving data synthesis, which produces multiple environments through task execution and generates diverse tasks and goal states. Complementing this, a state-anchor mechanism automatically annotates task-relevant UI elements in successful goal states as reference anchors. During RL training, these reference anchors provide accurate, scalable reward signals that substantially enhance efficiency. Extensive evaluations demonstrate that our framework achieves over 90% accuracy on offline trajectory verification and performs closest to rule-based methods. Furthermore, agents trained using our reward framework exhibit strong performance on both AndroidWorld and our constructed benchmark, establishing a scalable approach for GUI agent training.

cs.AI↗

arXiv2026

AssemPlanner: A Multi-Agent Based Task Planning Framework for Flexible Assembly System

In flexible assembly systems, existing task planning methods require a time-consuming configuration process by multiple experts to establish a production line for a new product. To address this challenge, we propose a multi-agent based task planning framework for flexible assembly systems, denoted as AssemPlanner. It takes tasks described in natural language as input, which are then converted into actionable sequential production operations. It comprises several specialized agents, including SchedAgent , KnowledgeAgent, LineBalanceAgent, and a scene graph. Within the proposed framework, SchedAgent serves as the central reasoning engine. Departing from traditional static pipelines, AssemPlanner utilizes a ReAct-based SchedAgent to adaptively adjust actions via multi-agent feedback. By observing the feedback from KnowledgeAgent, LineBalanceAgent, and the scene graph, it autonomously resolves complex industrial process constraints. To facilitate reproducibility, all code and datasets are released at https://github.com/chz332/Assemplanner.

cs.RO↗

arXiv2026

LEGO-MOF: Equivariant Latent Manipulation for Editable, Generative, and Optimizable MOF Design

Metal-organic frameworks (MOFs) are highly promising for carbon capture, yet navigating their vast design space remains challenging. Recent deep generative models enable de novo MOF design but primarily act as feed-forward structure generators. By heavily relying on predefined building block libraries and non-differentiable post-optimization, they fundamentally sever the information flow required for continuous structural editing. Here, we propose a target-driven generative framework focused on continuous structural manipulation. At its core is LinkerVAE, which maps discrete 3D chemical graphs into a continuous, SE(3)-equivariant latent space. This smooth manifold unlocks geometry-aware manipulations, including implicit chemical style transfer and zero-shot isoreticular expansion. Building upon this, we introduce a test-time optimization (TTO) strategy, utilizing an accurate surrogate model to continuously optimize the latent graphs of existing MOFs toward desired properties. This approach systematically enhances carbon capture performance, achieving a striking average relative boost of 147.5% in pure CO2 uptake while strictly preserving structural validity. Integrated with a latent diffusion model and rigid-body assembly for full MOF construction, our framework establishes a scalable, fully differentiable pathway for both the automated discovery, targeted optimization and editing of functional materials.

cs.LG↗

arXiv2026

MedLoc-R1: Performance-Aware Curriculum Reward Scheduling for GRPO-Based Medical Visual Grounding

Medical visual grounding serves as a crucial foundation for fine-grained multimodal reasoning and interpretable clinical decision support. Despite recent advances in reinforcement learning (RL) for grounding tasks, existing approaches such as Group Relative Policy Optimization~(GRPO) suffer from severe reward sparsity when directly applied to medical images, primarily due to the inherent difficulty of localizing small or ambiguous regions of interest, which is further exacerbated by the rigid and suboptimal nature of fixed IoU-based reward schemes in RL. This leads to vanishing policy gradients and stagnated optimization, particularly during early training. To address this challenge, we propose MedLoc-R1, a performance-aware reward scheduling framework that progressively tightens the reward criterion in accordance with model readiness. MedLoc-R1 introduces a sliding-window performance tracker and a multi-condition update rule that automatically adjust the reward schedule from dense, easily obtainable signals to stricter, fine-grained localization requirements, while preserving the favorable properties of GRPO without introducing auxiliary networks or additional gradient paths. Experiments on three medical visual grounding benchmarks demonstrate that MedLoc-R1 consistently improves both localization accuracy and training stability over GRPO-based baselines. Our framework offers a general, lightweight, and effective solution for RL-based grounding in high-stakes medical applications. Code \& checkpoints are available at \hyperlink{}{https://github.com/MembrAI/MedLoc-R1}.

cs.CV↗

arXiv2026

ATG-MoE: Autoregressive trajectory generation with mixture-of-experts for assembly skill learning

Flexible manufacturing requires robot systems that can adapt to constantly changing tasks, objects, and environments. However, traditional robot programming is labor-intensive and inflexible, while existing learning-based assembly methods often suffer from weak positional generalization, complex multi-stage designs, and limited multi-skill integration capability. To address these issues, this paper proposes ATG-MoE, an end-to-end autoregressive trajectory generation method with mixture of experts for assembly skill learning from demonstration. The proposed method establishes a closed-loop mapping from multi-modal inputs, including RGB-D observations, natural language instructions, and robot proprioception to manipulation trajectories. It integrates multi-modal feature fusion for scene and task understanding, autoregressive sequence modeling for temporally coherent trajectory generation, and a mixture-of-experts architecture for unified multi-skill learning. In contrast to conventional methods that separate visual perception and control or train different skills independently, ATG-MoE directly incorporates visual information into trajectory generation and supports efficient multi-skill integration within a single model. We train and evaluate the proposed method on eight representative assembly skills from a pressure-reducing valve assembly task. Experimental results show that ATG-MoE achieves strong overall performance in simulation, with an average grasp success rate of 96.3% and an average overall success rate of 91.8%, while also demonstrating strong generalization and effective multi-skill integration. Real-world experiments further verify its practicality for multi-skill industrial assembly. The project page can be found at https://hwh23.github.io/ATG-MoE

cs.RO↗

arXiv2026

ForeHOI: Feed-forward 3D Object Reconstruction from Daily Hand-Object Interaction Videos

The ubiquity of monocular videos capturing daily hand-object interactions presents a valuable resource for embodied intelligence. While 3D hand reconstruction from in-the-wild videos has seen significant progress, reconstructing the involved objects remains challenging due to severe occlusions and the complex, coupled motion of the camera, hands, and object. In this paper, we introduce ForeHOI, a novel feed-forward model that directly reconstructs 3D object geometry from monocular hand-object interaction videos within one minute of inference time, eliminating the need for any pre-processing steps. Our key insight is that, the joint prediction of 2D mask inpainting and 3D shape completion in a feed-forward framework can effectively address the problem of severe occlusion in monocular hand-held object videos, thereby achieving results that outperform the performance of optimization-based methods. The information exchanges between the 2D and 3D shape completion boosts the overall reconstruction quality, enabling the framework to effectively handle severe hand-object occlusion. Furthermore, to support the training of our model, we contribute the first large-scale, high-fidelity synthetic dataset of hand-object interactions with comprehensive annotations. Extensive experiments demonstrate that ForeHOI achieves state-of-the-art performance in object reconstruction, significantly outperforming previous methods with around a 100x speedup. Code and data are available at: https://github.com/Tao-11-chen/ForeHOI.

cs.CV↗

arXiv2025

AssemMate: Graph-Based LLM for Robotic Assembly Assistance

Large Language Model (LLM)-based robotic assembly assistance has gained significant research attention. It requires the injection of domain-specific knowledge to guide the assembly process through natural language interaction with humans. Despite some progress, existing methods represent knowledge in the form of natural language text. Due to the long context and redundant content, they struggle to meet the robots' requirements for real-time and precise reasoning. In order to bridge this gap, we present AssemMate, which utilizes the graph\textemdash a concise and accurate form of knowledge representation\textemdash as input. This graph-based LLM enables knowledge graph question answering (KGQA), supporting human-robot interaction and assembly task planning for specific products. Beyond interactive QA, AssemMate also supports sensing stacked scenes and executing grasping to assist with assembly. Specifically, a self-supervised Graph Convolutional Network (GCN) encodes knowledge graph entities and relations into a latent space and aligns them with LLM's representation, enabling the LLM to understand graph information. In addition, a vision-enhanced strategy is employed to address stacked scenes in grasping. Through training and evaluation, AssemMate outperforms existing methods, achieving 6.4\% higher accuracy, 3 times faster inference, and 28 times shorter context length, while demonstrating strong generalization ability on random graphs. And our approach further demonstrates superiority through robotic grasping experiments in both simulated and real-world settings. More details can be found on the project page: https://github.com/cristina304/AssemMate.git

cs.RO↗

arXiv2025

Embodied intelligent industrial robotics: Framework and techniques

The combination of embodied intelligence and robots has great prospects and is becoming increasingly common. In order to work more efficiently, accurately, reliably, and safely in industrial scenarios, robots should have at least general knowledge, working-environment knowledge, and operating-object knowledge. These pose significant challenges to existing embodied intelligent robotics (EIR) techniques. Thus, this paper first briefly reviews the history of industrial robotics and analyzes the limitations of mainstream EIR frameworks. Then, a new knowledge-driven technical framework of embodied intelligent industrial robotics (EIIR) is proposed for various industrial environments. It has five modules: a world model, a high-level task planner, a low-level skill controller, a simulator, and a physical system. The development of techniques related to each module are also thoroughly reviewed, and recent progress regarding their adaption to industrial applications are discussed. A case study of real-world assembly system is given to demonstrate the newly proposed EIIR framework's applicability and potentiality. Finally, the key challenges that EIIR encounters in industrial scenarios are summarized and future research directions are suggested. The authors believe that EIIR technology is shaping the next generation of industrial robotics and EIIR-based industrial systems supply a new technological paradigm for intelligent manufacturing. It is expected that this review could serve as a valuable reference for scholars and engineers that are interested in industrial embodied intelligence. Together, scholars can use this research to drive their rapid advancement and application of EIIR techniques. The authors would continue to track and contribute new studies in the project page https://github.com/jackyzengl/EIIR

cs.RO↗

arXiv2024

Prospects for Systematic Planetary Nebulae Detection with the Census of the Local Universe Narrowband Survey

We investigate the efficacy of a systematic planetary nebula (PN) search in the Census of the Local Universe (CLU) narrowband (H$\alpha$) survey that covers a considerably larger sky region of above declination $-20^\circ$ than most previous surveys. Using PNe observed by the Isaac Newton Telescope Photometric H$\alpha$ Survey (IPHAS) as validation, we are able to visually recover 432 out of 441 cataloged PNe (98\%) within the CLU dataset, with 5 sources having unusable CLU images and 4 missed due to limitations of imaging quality. Moreover, the reference PNe are conventionally divided into three PN classes in decreasing order of identification confidence given their spectra and morphologies. We record consistently high recovery rate across all classes: 95\% of True, 71\% of Likely, and 81\% of Possible sources are readily recovered. To further demonstrate the ability of CLU to find new PNe, we undertake a preliminary search of compact PNe within a sub-region of the validation catalog, mainly utilizing the significance of narrow-band colors ($\Sigma$) as a metric for identification. In a $200\,\rm deg^2$ region, we search the CLU source catalog and find 31 PN candidates after automated and visual scrutiny, of which 12 are new sources not appearing in previous studies. As a demonstration of our ongoing follow-up campaign, we present medium-resolution optical spectra of six candidates and notice that four of them show emission signatures characteristic of confirmed PNe. As we refine our selection methods, CLU promises to provide a systematic catalog of PNe spanning $2/3$ of the sky.

astro-ph.SR↗

arXiv2024

Efficient Sparse Attention needs Adaptive Token Release

In recent years, Large Language Models (LLMs) have demonstrated remarkable capabilities across a wide array of text-centric tasks. However, their `large' scale introduces significant computational and storage challenges, particularly in managing the key-value states of the transformer, which limits their wider applicability. Therefore, we propose to adaptively release resources from caches and rebuild the necessary key-value states. Particularly, we accomplish this by a lightweight controller module to approximate an ideal top-$K$ sparse attention. This module retains the tokens with the highest top-$K$ attention weights and simultaneously rebuilds the discarded but necessary tokens, which may become essential for future decoding. Comprehensive experiments in natural language generation and modeling reveal that our method is not only competitive with full attention in terms of performance but also achieves a significant throughput improvement of up to 221.8%. The code for replication is available on the https://github.com/WHUIR/ADORE.

cs.CL↗

arXiv2024

Long-rising Type II Supernovae in the Zwicky Transient Facility Census of the Local Universe

SN 1987A was an unusual hydrogen-rich core-collapse supernova originating from a blue supergiant star. Similar blue supergiant explosions remain a small family of events, and are broadly characterized by their long rises to peak. The Zwicky Transient Facility (ZTF) Census of the Local Universe (CLU) experiment aims to construct a spectroscopically complete sample of transients occurring in galaxies from the CLU galaxy catalog. We identify 13 long-rising (>40 days) Type II supernovae from the volume-limited CLU experiment during a 3.5 year period from June 2018 to December 2021, approximately doubling the previously known number of these events. We present photometric and spectroscopic data of these 13 events, finding peak r-band absolute magnitudes ranging from -15.6 to -17.5 mag and the tentative detection of Ba II lines in 9 events. Using our CLU sample of events, we derive a long-rising Type II supernova rate of $1.37^{+0.26}_{-0.30}\times10^{-6}$ Mpc$^{-3}$ yr$^{-1}$, $\approx$1.4% of the total core-collapse supernova rate. This is the first volumetric rate of these events estimated from a large, systematic, volume-limited experiment.

astro-ph.HE↗

arXiv2024

Collapsars as Sites of r-process Nucleosynthesis: Systematic Near-Infrared Follow-up of Type Ic-BL Supernovae

One of the open questions following the discovery of GW170817 is whether neutron star mergers are the only astrophysical sites capable of producing $r$-process elements. Simulations have shown that 0.01-0.1M$_\odot$ of $r$-process material could be generated in the outflows originating from the accretion disk surrounding the rapidly rotating black hole that forms as a remnant to both neutron star mergers and collapsing massive stars associated with long-duration gamma-ray bursts (collapsars). The hallmark signature of $r$-process nucleosynthesis in the binary neutron star merger GW170817 was its long-lasting near-infrared emission, thus motivating a systematic photometric study of the light curves of broadlined stripped-envelope (Ic-BL) supernovae (SNe) associated with collapsars. We present the first systematic study of 25 SNe Ic-BL -- including 18 observed with the Zwicky Transient Facility and 7 from the literature -- in the optical/near-infrared bands to determine what quantity of $r$-process material, if any, is synthesized in these explosions. Using semi-analytic models designed to account for $r$-process production in SNe Ic-BL, we perform light curve fitting to derive constraints on the $r$-process mass for these SNe. We also perform independent light curve fits to models without $r$-process. We find that the $r$-process-free models are a better fit to the light curves of the objects in our sample. Thus we find no compelling evidence of $r$-process enrichment in any of our objects. Further high-cadence infrared photometric studies and nebular spectroscopic analysis would be sensitive to smaller quantities of $r$-process ejecta mass or indicate whether all collapsars are completely devoid of $r$-process nucleosynthesis.

astro-ph.HE↗

arXiv2023

GiPH: Generalizable Placement Learning for Adaptive Heterogeneous Computing

Careful placement of a computational application within a target device cluster is critical for achieving low application completion time. The problem is challenging due to its NP-hardness and combinatorial nature. In recent years, learning-based approaches have been proposed to learn a placement policy that can be applied to unseen applications, motivated by the problem of placing a neural network across cloud servers. These approaches, however, generally assume the device cluster is fixed, which is not the case in mobile or edge computing settings, where heterogeneous devices move in and out of range for a particular application. We propose a new learning approach called GiPH, which learns policies that generalize to dynamic device clusters via 1) a novel graph representation gpNet that efficiently encodes the information needed for choosing a good placement, and 2) a scalable graph neural network (GNN) that learns a summary of the gpNet information. GiPH turns the placement problem into that of finding a sequence of placement improvements, learning a policy for selecting this sequence that scales to problems of arbitrary size. We evaluate GiPH with a wide range of task graphs and device clusters and show that our learned policy rapidly find good placements for new problem instances. GiPH finds placements with up to 30.5% lower completion times, searching up to 3X faster than other search-based placement policies.

cs.LG↗

arXiv2022

Volumetric rates of Luminous Red Novae and Intermediate Luminosity Red Transients with the Zwicky Transient Facility

Luminous red novae (LRNe) are transients characterized by low luminosities and expansion velocities, and are associated with mergers or common envelope ejections in stellar binaries. Intermediate-luminosity red transients (ILRTs) are an observationally similar class with unknown origins, but generally believed to either be electron capture supernovae (ECSN) in super-AGB stars, or outbursts in dusty luminous blue variables (LBVs). In this paper, we present a systematic sample of 8 LRNe and 8 ILRTs detected as part of the Census of the Local Universe (CLU) experiment on the Zwicky Transient Facility (ZTF). The CLU experiment spectroscopically classifies ZTF transients associated with nearby ($<150$ Mpc) galaxies, achieving 80% completeness for m$_{r}<20$\,mag. Using the ZTF-CLU sample, we derive the first systematic LRNe volumetric-rate of 7.8$^{+6.5}_{-3.7}\times10^{-5}$ Mpc$^{-3}$ yr$^{-1}$ in the luminosity range $-16\leq$M$_{\rm{r}}$$\leq -11$ mag. We find that in this luminosity range, the LRN rate scales as dN/dL $\propto L^{-2.5\pm0.3}$ - significantly steeper than the previously derived scaling of $L^{-1.4\pm0.3}$ for lower luminosity LRNe (M$_{V}\geq-10$). The steeper power law for LRNe at high luminosities is consistent with the massive merger rates predicted by binary population synthesis models. We find that the rates of the brightest LRNe (M$_{r}\leq-13$ mag) are consistent with a significant fraction of them being progenitors of double compact objects (DCOs) that merge within a Hubble time. For ILRTs, we derive a volumetric rate of $2.6^{+1.8}_{-1.4}\times10^{-6}$ Mpc$^{-3}$yr$^{-1}$ for M$_{\rm{r}}\leq-13.5$, that scales as dN/dL $\propto L^{-2.5\pm0.5}$. This rate is $\approx1-5\%$ of the local core-collapse supernova rate, and is consistent with theoretical ECSN rate estimates.

astro-ph.HE↗

arXiv2022

Faithful Explanations for Deep Graph Models

This paper studies faithful explanations for Graph Neural Networks (GNNs). First, we provide a new and general method for formally characterizing the faithfulness of explanations for GNNs. It applies to existing explanation methods, including feature attributions and subgraph explanations. Second, our analytical and empirical results demonstrate that feature attribution methods cannot capture the nonlinear effect of edge features, while existing subgraph explanation methods are not faithful. Third, we introduce \emph{k-hop Explanation with a Convolutional Core} (KEC), a new explanation method that provably maximizes faithfulness to the original GNN by leveraging information about the graph structure in its adjacency matrix and its \emph{k-th} power. Lastly, our empirical results over both synthetic and real-world datasets for classification and anomaly detection tasks with GNNs demonstrate the effectiveness of our approach.

cs.LG↗