arXiv subjects

Haoyun Li

Publications and source records attributed to Haoyun Li.

18 recordsLinked to original sources

arXiv2026

ComAgent: Multi-LLM based Agentic AI Empowered Intelligent Wireless Networks

Emerging 6G networks rely on complex cross-layer optimization, yet manually translating high-level intents into mathematical formulations remains a bottleneck. While Large Language Models (LLMs) offer promise, monolithic approaches often lack sufficient domain grounding, constraint awareness, and verification capabilities. To address this, we present ComAgent, a multi-LLM agentic AI framework. ComAgent employs a closed-loop Perception-Planning-Action-Reflection cycle, coordinating specialized agents for literature search, coding, and scoring to autonomously generate solver-ready formulations and reproducible simulations. By iteratively decomposing problems and self-correcting errors, the framework effectively bridges the gap between user intent and execution. Evaluations demonstrate that ComAgent achieves expert-comparable performance in complex beamforming optimization and outperforms monolithic LLMs across diverse wireless tasks, highlighting its potential for automating design in emerging wireless networks.

cs.AI↗

arXiv2025

SwiftVLA: Unlocking Spatiotemporal Dynamics for Lightweight VLA Models at Minimal Overhead

Vision-Language-Action (VLA) models built on pretrained Vision-Language Models (VLMs) show strong potential but are limited in practicality due to their large parameter counts. To mitigate this issue, using a lightweight VLM has been explored, but it compromises spatiotemporal reasoning. Although some methods suggest that incorporating additional 3D inputs can help, they usually rely on large VLMs to fuse 3D and 2D inputs and still lack temporal understanding. Therefore, we propose SwiftVLA, an architecture that enhances a compact model with 4D understanding while preserving design efficiency. Specifically, our approach features a pretrained 4D visual geometry transformer with a temporal cache that extracts 4D features from 2D images. Then, to enhance the VLM's ability to exploit both 2D images and 4D features, we introduce Fusion Tokens, a set of learnable tokens trained with a future prediction objective to generate unified representations for action generation. Finally, we introduce a mask-and-reconstruct strategy that masks 4D inputs to the VLM and trains the VLA to reconstruct them, enabling the VLM to learn effective 4D representations and allowing the 4D branch to be dropped at inference with minimal performance loss. Experiments in real and simulated environments show that SwiftVLA outperforms lightweight baselines and rivals VLAs up to 7 times larger, achieving comparable performance on edge devices while being 18 times faster and reducing memory footprint by 12 times.

cs.CV↗

arXiv2025

GigaWorld-0: World Models as Data Engine to Empower Embodied AI

World models are emerging as a foundational paradigm for scalable, data-efficient embodied AI. In this work, we present GigaWorld-0, a unified world model framework designed explicitly as a data engine for Vision-Language-Action (VLA) learning. GigaWorld-0 integrates two synergistic components: GigaWorld-0-Video, which leverages large-scale video generation to produce diverse, texture-rich, and temporally coherent embodied sequences under fine-grained control of appearance, camera viewpoint, and action semantics; and GigaWorld-0-3D, which combines 3D generative modeling, 3D Gaussian Splatting reconstruction, physically differentiable system identification, and executable motion planning to ensure geometric consistency and physical realism. Their joint optimization enables the scalable synthesis of embodied interaction data that is visually compelling, spatially coherent, physically plausible, and instruction-aligned. Training at scale is made feasible through our efficient GigaTrain framework, which exploits FP8-precision and sparse attention to drastically reduce memory and compute requirements. We conduct comprehensive evaluations showing that GigaWorld-0 generates high-quality, diverse, and controllable data across multiple dimensions. Critically, VLA model (e.g., GigaBrain-0) trained on GigaWorld-0-generated data achieve strong real-world performance, significantly improving generalization and task success on physical robots without any real-world interaction during training.

cs.CV↗

arXiv2025

GigaBrain-0: A World Model-Powered Vision-Language-Action Model

Training Vision-Language-Action (VLA) models for generalist robots typically requires large-scale real-world robot data, which is expensive and time-consuming to collect. The inefficiency of physical data collection severely limits the scalability, and generalization capacity of current VLA systems. To address this challenge, we introduce GigaBrain-0, a novel VLA foundation model empowered by world model-generated data (e.g., video generation, real2real transfer, human transfer, view transfer, sim2real transfer data). By leveraging world models to generate diverse data at scale, GigaBrain-0 significantly reduces reliance on real robot data while improving cross-task generalization. Our approach further improves policy robustness through RGBD input modeling and embodied Chain-of-Thought (CoT) supervision, enabling the model to reason about spatial geometry, object states, and long-horizon dependencies during task execution. This leads to substantial gains in real-world performance on dexterous, long-horizon, and mobile manipulation tasks. Extensive experiments demonstrate that GigaBrain-0 achieves superior generalization across variations in appearances (e.g., textures, colors), object placements, and camera viewpoints. Additionally, we present GigaBrain-0-Small, an optimized lightweight variant designed to run efficiently on devices such as the NVIDIA Jetson AGX Orin.

cs.RO↗

arXiv2025

DriveGen3D: Boosting Feed-Forward Driving Scene Generation with Efficient Video Diffusion

We present DriveGen3D, a novel framework for generating high-quality and highly controllable dynamic 3D driving scenes that addresses critical limitations in existing methodologies. Current approaches to driving scene synthesis either suffer from prohibitive computational demands for extended temporal generation, focus exclusively on prolonged video synthesis without 3D representation, or restrict themselves to static single-scene reconstruction. Our work bridges this methodological gap by integrating accelerated long-term video generation with large-scale dynamic scene reconstruction through multimodal conditional control. DriveGen3D introduces a unified pipeline consisting of two specialized components: FastDrive-DiT, an efficient video diffusion transformer for high-resolution, temporally coherent video synthesis under text and Bird's-Eye-View (BEV) layout guidance; and FastRecon3D, a feed-forward module that rapidly builds 3D Gaussian representations across time, ensuring spatial-temporal consistency. DriveGen3D enable the generation of long driving videos (up to $800\times424$ at $12$ FPS) and corresponding 3D scenes, achieving state-of-the-art results while maintaining efficiency.

cs.CV↗

arXiv2025

MimicDreamer: Aligning Human and Robot Demonstrations for Scalable VLA Training

Vision Language Action (VLA) models derive their generalization capability from diverse training data, yet collecting embodied robot interaction data remains prohibitively expensive. In contrast, human demonstration videos are far more scalable and cost-efficient to collect, and recent studies confirm their effectiveness in training VLA models. However, a significant domain gap persists between human videos and robot-executed videos, including unstable camera viewpoints, visual discrepancies between human hands and robotic arms, and differences in motion dynamics. To bridge this gap, we propose MimicDreamer, a framework that turns fast, low-cost human demonstrations into robot-usable supervision by jointly aligning vision, viewpoint, and actions to directly support policy training. For visual alignment, we propose H2R Aligner, a video diffusion model that generates high-fidelity robot demonstration videos by transferring motion from human manipulation footage. For viewpoint stabilization, EgoStabilizer is proposed, which canonicalizes egocentric videos via homography and inpaints occlusions and distortions caused by warping. For action alignment, we map human hand trajectories to the robot frame and apply a constrained inverse kinematics solver to produce feasible, low-jitter joint commands with accurate pose tracking. Empirically, VLA models trained purely on our synthesized human-to-robot videos achieve few-shot execution on real robots. Moreover, scaling training with human data significantly boosts performance compared to models trained solely on real robot data; our approach improves the average success rate by 14.7\% across six representative manipulation tasks.

cs.RO↗

arXiv2025

WonderFree: Enhancing Novel View Quality and Cross-View Consistency for 3D Scene Exploration

Interactive 3D scene generation from a single image has gained significant attention due to its potential to create immersive virtual worlds. However, a key challenge in current 3D generation methods is the limited explorability, which cannot render high-quality images during larger maneuvers beyond the original viewpoint, particularly when attempting to move forward into unseen areas. To address this challenge, we propose WonderFree, the first model that enables users to interactively generate 3D worlds with the freedom to explore from arbitrary angles and directions. Specifically, we decouple this challenge into two key subproblems: novel view quality, which addresses visual artifacts and floating issues in novel views, and cross-view consistency, which ensures spatial consistency across different viewpoints. To enhance rendering quality in novel views, we introduce WorldRestorer, a data-driven video restoration model designed to eliminate floaters and artifacts. In addition, a data collection pipeline is presented to automatically gather training data for WorldRestorer, ensuring it can handle scenes with varying styles needed for 3D scene generation. Furthermore, to improve cross-view consistency, we propose ConsistView, a multi-view joint restoration mechanism that simultaneously restores multiple perspectives while maintaining spatiotemporal coherence. Experimental results demonstrate that WonderFree not only enhances rendering quality across diverse viewpoints but also significantly improves global coherence and consistency. These improvements are confirmed by CLIP-based metrics and a user study showing a 77.20% preference for WonderFree over WonderWorld enabling a seamless and immersive 3D exploration experience. The code, model, and data will be publicly available.

cs.CV↗

arXiv2025

Motion-R1: Enhancing Motion Generation with Decomposed Chain-of-Thought and RL Binding

Text-to-Motion generation has become a fundamental task in human-machine interaction, enabling the synthesis of realistic human motions from natural language descriptions. Although recent advances in large language models and reinforcement learning have contributed to high-quality motion generation, two major challenges remain. Existing approaches often fail to capture the temporal and causal complexities inherent in natural language, leading to oversimplified or incoherent motions. Additionally, RL-based methods are frequently overly complex, hindering their scalability and adaptability across various motion generation tasks. To address these challenges, we propose Motion-R1, a novel framework that combines decomposed Chain-of-Thought reasoning with reinforcement learning to enhance both the quality and interpretability of generated motions. Specifically, we introduce the Decomposed CoT Data Engine, which leverages an automated pipeline to synthesize high-quality reasoning data, allowing the model to better capture the temporal dependencies and causal relationships of human motion. We also propose RL Binding, a reinforcement learning strategy that incorporates multi-modal text-motion alignment into the RL reward function, guiding the model to produce motions that are both semantically accurate and motionally realistic. Extensive experiments across benchmark datasets demonstrate that Motion-R1 achieves state-of-the-art performance, with a 3.5% improvement in MM-Dist on HumanML3D and improvements in R-Precision and FID on KIT-ML and BABEL, surpassing existing methods across key metrics and highlighting its superior capability in handling complex motion generation tasks. Project page: https://motion-r1.github.io/.

cs.CV↗

arXiv2025

MIMO Pinching-Antenna-Aided SWIPT

Pinching-antenna systems (PASS) have recently emerged as a promising technology for improving wireless communications by establishing or strengthening reliable line-of-sight (LoS) links by adjusting the positions of pinching antennas (PAs). Motivated by these benefits, we propose a novel PASS-aided multi-input multi-output (MIMO) system for simultaneous wireless information and power transfer (SWIPT), where the PASS are equipped with multiple waveguides to provide information transmission and wireless power transfer (WPT) for several multiple antenna information decoding receivers (IDRs), and energy harvesting receivers (EHRs), respectively. Based on the system, we consider maximizing the sum-rate of all IDRs while guaranteeing the minimum harvested energy of each EHR by jointly optimizing the pinching beamforming and the PA positions. To solve this highly non-convex problem, we iteratively optimize the pinching beamforming based on a weighted minimum mean-squared-error (WMMSE) method and update the PA positions with a Gauss-Seidel-based approach in an alternating optimization (AO) framework. Numerical results verify the significant superiority of the PASS compared with conventional designs.

cs.IT↗

arXiv2025

Joint User Association and Beamforming Design for ISAC Networks with Large Language Models

Integrated sensing and communication (ISAC) has been envisioned to play a more important role in future wireless networks. However, the design of ISAC networks is challenging, especially when there are multiple communication and sensing (C\&S) nodes and multiple sensing targets. We investigate a multi-base station (BS) ISAC network in which multiple BSs equipped with multiple antennas simultaneously provide C\&S services for multiple ground communication users (CUs) and targets. To enhance the overall performance of C\&S, we formulate a joint user association (UA) and multi-BS transmit beamforming optimization problem with the objective of maximizing the total sum rate of all CUs while ensuring both the minimum target detection and parameter estimation requirements. To efficiently solve the highly non-convex mixed integer nonlinear programming (MINLP) optimization problem, we propose an alternating optimization (AO)-based algorithm that decomposes the problem into two sub-problems, i.e., UA optimization and multi-BS transmit beamforming optimization. Inspired by large language models (LLMs) for prediction and inference, we propose a unified framework integrating LLMs with convex-based optimization methods. First, we propose a comprehensive design of prompt engineering, including few-shot, chain of thought, and self-reflection techniques to guide LLMs in solving the binary integer programming UA optimization problem. Second, we utilize convex-based optimization methods to handle the non-convex beamforming optimization problem based on fractional programming (FP), majorization minimization (MM), and the alternating direction method of multipliers (ADMM) with an optimized UA from LLMs. Numerical results demonstrate that our proposed LLM-enabled AO-based algorithm achieves fast convergence and near upper-bound performance with the GPT-o1 model, outperforming various benchmark schemes.

cs.IT↗

arXiv2025

Pinching-Antenna Systems (PASS) Aided Over-the-air Computation

Over-the-air computation (AirComp) enables fast data aggregation for edge intelligence applications. However the performance of AirComp can be severely degraded by channel misalignments. Pinching antenna systems (PASS) have recently emerged as a promising solution for physically reshaping favorable wireless channels to reduce misalignments and thus AirComp errors, via low-cost, fully passive, and highly reconfigurable antenna deployment. Motivated by these benefits, we propose a novel PASS-aided AirComp system that introduces new design degrees of freedom through flexible pinching antenna (PA) placement. To improve performance, we consider a mean squared error (MSE) minimization problem by jointly optimizing the PA position, transmit power, and decoding vector. To solve this highly non-convex problem, we propose an alternating optimization based framework with Gauss-Seidel based PA position updates. Simulation results show that our proposed joint PA position and communication design significantly outperforms various benchmark schemes in AirComp accuracy.

cs.IT↗

arXiv2025

Well2Flow: Reconstruction of reservoir states from sparse wells using score-based generative models

This study investigates the use of score-based generative models for reservoir simulation, with a focus on reconstructing spatially varying permeability and saturation fields in saline aquifers, inferred from sparse observations at two well locations. By modeling the joint distribution of permeability and saturation derived from high-fidelity reservoir simulations, the proposed neural network is trained to learn the complex spatiotemporal dynamics governing multiphase fluid flow in porous media. During inference, the framework effectively reconstructs both permeability and saturation fields by conditioning on sparse vertical profiles extracted from well log data. This approach introduces a novel methodology for incorporating physical constraints and well log guidance into generative models, significantly enhancing the accuracy and physical plausibility of the reconstructed subsurface states. Furthermore, the framework demonstrates strong generalization capabilities across varying geological scenarios, highlighting its potential for practical deployment in data-scarce reservoir management tasks.

cs.LG↗

arXiv2025

WonderTurbo: Generating Interactive 3D World in 0.72 Seconds

Interactive 3D generation is gaining momentum and capturing extensive attention for its potential to create immersive virtual experiences. However, a critical challenge in current 3D generation technologies lies in achieving real-time interactivity. To address this issue, we introduce WonderTurbo, the first real-time interactive 3D scene generation framework capable of generating novel perspectives of 3D scenes within 0.72 seconds. Specifically, WonderTurbo accelerates both geometric and appearance modeling in 3D scene generation. In terms of geometry, we propose StepSplat, an innovative method that constructs efficient 3D geometric representations through dynamic updates, each taking only 0.26 seconds. Additionally, we design QuickDepth, a lightweight depth completion module that provides consistent depth input for StepSplat, further enhancing geometric accuracy. For appearance modeling, we develop FastPaint, a 2-steps diffusion model tailored for instant inpainting, which focuses on maintaining spatial appearance consistency. Experimental results demonstrate that WonderTurbo achieves a remarkable 15X speedup compared to baseline methods, while preserving excellent spatial consistency and delivering high-quality output.

cs.CV↗

arXiv2024

Large Language Model Based Multi-Objective Optimization for Integrated Sensing and Communications in UAV Networks

This letter investigates an unmanned aerial vehicle (UAV) network with integrated sensing and communication (ISAC) systems, where multiple UAVs simultaneously sense the locations of ground users and provide communication services with radars. To find the trade-off between communication and sensing (C\&S) in the system, we formulate a multi-objective optimization problem (MOP) to maximize the total network utility and the localization Cram\'er-Rao bounds (CRB) of ground users, which jointly optimizes the deployment and power control of UAVs. Inspired by the huge potential of large language models (LLM) for prediction and inference, we propose an LLM-enabled decomposition-based multi-objective evolutionary algorithm (LEDMA) for solving the highly non-convex MOP. We first adopt a decomposition-based scheme to decompose the MOP into a series of optimization sub-problems. We second integrate LLMs as black-box search operators with MOP-specifically designed prompt engineering into the framework of MOEA to solve optimization sub-problems simultaneously. Numerical results demonstrate that the proposed LEDMA can find the clear trade-off between C\&S and outperforms baseline MOEAs in terms of obtained Pareto fronts and convergence.

cs.IT↗

arXiv2024

A Digital Twin for Geological Carbon Storage with Controlled Injectivity

We present an uncertainty-aware Digital Twin (DT) for geologic carbon storage (GCS), capable of handling multimodal time-lapse data and controlling CO2 injectivity to mitigate reservoir fracturing risks. In GCS, DT represents virtual replicas of subsurface systems that incorporate real-time data and advanced generative Artificial Intelligence (genAI) techniques, including neural posterior density estimation via simulation-based inference and sequential Bayesian inference. These methods enable the effective monitoring and control of CO2 storage projects, addressing challenges such as subsurface complexity, operational optimization, and risk mitigation. By integrating diverse monitoring data, e.g., geophysical well observations and imaged seismic, DT can bridge the gaps between seemingly distinct fields like geophysics and reservoir engineering. In addition, the recent advancements in genAI also facilitate DT with principled uncertainty quantification. Through recursive training and inference, DT utilizes simulated current state samples, e.g., CO2 saturation, paired with corresponding geophysical field observations to train its neural networks and enable posterior sampling upon receiving new field data. However, it lacks decision-making and control capabilities, which is necessary for full DT functionality. This study aims to demonstrate how DT can inform decision-making processes to prevent risks such as cap rock fracturing during CO2 storage operations.

physics.geo-ph↗

arXiv2023

Surrogate-based Autotuning for Randomized Sketching Algorithms in Regression Problems

Algorithms from Randomized Numerical Linear Algebra (RandNLA) are known to be effective in handling high-dimensional computational problems, providing high-quality empirical performance as well as strong probabilistic guarantees. However, their practical application is complicated by the fact that the user needs to set various algorithm-specific tuning parameters which are different than those used in traditional NLA. This paper demonstrates how a surrogate-based autotuning approach can be used to address fundamental problems of parameter selection in RandNLA algorithms. In particular, we provide a detailed investigation of surrogate-based autotuning for sketch-and-precondition (SAP) based randomized least squares methods, which have been one of the great success stories in modern RandNLA. Empirical results show that our surrogate-based autotuning approach can achieve near-optimal performance with much less tuning cost than a random search (up to about 4x fewer trials of different parameter configurations). Moreover, while our experiments focus on least squares, our results demonstrate a general-purpose autotuning pipeline applicable to any kind of RandNLA algorithm.

cs.LG↗

arXiv2022

Channel Knowledge Map (CKM)-Assisted Multi-UAV Wireless Network: CKM Construction and UAV Placement

Channel knowledge map (CKM) has recently emerged to facilitate the placement and trajectory optimization for unmanned aerial vehicle (UAV) communications. This paper investigates a CKM-assisted multi-UAV wireless network, by focusing on the construction and utilization of CKMs for multi-UAV placement optimization. First, we consider the CKM construction problem when data measurements for only a limited number of points are available. Towards this end, we exploit a data-driven interpolation technique to construct CKMs to characterize the signal propagation environments. Next, we study the multi-UAV placement optimization problem by utilizing the constructed CKMs, in which the multiple UAVs aim to optimize their placement locations to maximize the weighted sum rate with their respectively associated ground base stations (GBSs). However, the rate function based on the CKMs is generally non-differentiable. To tackle this issue, we propose a novel iterative algorithm based on derivative-free optimization, in which a series of quadratic functions are iteratively constructed to approximate the objective function under a set of interpolation conditions, and accordingly, the UAVs' placement locations are updated by maximizing the approximate function subject to a trust region constraint. Finally, numerical results are presented to validate the proposed design achieves near-optimal performance, but with much lower implementation complexity.

cs.IT↗

arXiv2022

Derivative-Free Placement Optimization for Multi-UAV Wireless Networks with Channel Knowledge Map

This paper studies a multi-UAV wireless network, in which multiple UAV users share the same spectrum to send individual messages to their respectively associated ground base stations (GBSs). The UAV users aim to optimize their locations to maximize the weighted sum rate. While most existing work considers simplified line-of-sight (LoS) or statistic air-to-ground (A2G) channel models, we exploit the location-specific channel knowledge map (CKM) to enhance the placement performance in practice. However, as the CKMs normally contain discrete site- and location-specific channel data without analytic model functions, the corresponding weighted sum rate function becomes non-differentiable in general. In this case, conventional optimization techniques relying on function derivatives are inapplicable to solve the resultant placement optimization problem. To address this issue, we propose a novel iterative algorithm based on the derivative-free optimization. In each iteration, we first construct a quadratic function to approximate the non-differentiable weighted sum rate under a set of interpolation conditions, and then update the UAVs' placement locations by maximizing the approximate quadratic function subject to a trust region constraint. Numerical results show the convergence of the proposed algorithm. It is also shown that the proposed algorithm achieves a weighted sum rate close to the optimal design based on exhaustive search with much lower implementation complexity, and it significantly outperforms the conventional optimization method based on simplified LoS channel models and the heuristic design with each UAV hovering above its associated GBS.

cs.IT↗