arXiv subjects

Mengmeng Liu

Publications and source records attributed to Mengmeng Liu.

At least 19 recordsLinked to original sources

arXiv2026

Analysis and Approximation of Stochastic Multiscale Subdiffusion Driven by Fractional Gaussian Noise

This paper investigates a stochastic multiscale subdiffusion model driven by fractional Gaussian noise, where the multiscale Abel kernel with variable exponent $\alpha(t)\in(0,1)$ is used to capture multiscale and crossover behavior in anomalous diffusion. The main difficulties of this model lie in the complexity of the multiscale Abel kernel (e.g. non-monotonicity and non-coercivity) and the low regularity caused by the noise. Concerning these issues, we prove the well-posedness and regularity of the mild solutions by means of solution operator approach and a perturbation technique for multiscale Abel kernel. Then both the semidiscrete-in-time and fully-discrete numerical schemes are proposed and analyzed under the low-regularity numerical analysis framework, with proved temporal and spatial convergence rates. Numerical experiments are presented to substantiate the theoretical results.

math.NA↗

arXiv2026

UNIVERSE: Unified Video Action Models for Autonomous Driving with Flexible Mask-Modulated Modality Generation

World Action Models (WAMs) have shown strong potential for improving action generalization in autonomous driving by using future video prediction as dense supervision for scene dynamics and temporal causality. However, it remains unclear which architecture better transfers video-modeling benefits to trajectory generation. Existing cascaded or dual-DiT designs separate video imagination from action prediction, weakening the transfer of video-learned world dynamics to the trajectory branch: the action model may still overfit dataset-specific driving priors, while the video model only indirectly regularizes planning. We propose UNIVERSE, a unified video-action model built upon a single mask-modulated Diffusion Transformer. By co-training future video latents and ego-trajectory tokens within shared generative parameters, UNIVERSE allows dense video supervision to directly shape trajectory denoising, leading to stronger cross-domain action generalization. To ensure causal validity and efficient deployment, we introduce a Modality-Decoupling Visibility Mask, which shares historical context across modalities while blocking mutual attention between future video and trajectory tokens. This prevents future-target leakage and enables trajectory-only inference by removing future-video denoising at test time, achieving a $4.3\times$ speedup over joint video-action rollout while maintaining comparable planning accuracy. The same model also supports video-only and joint video-action rollouts. Experiments show that UNIVERSE achieves 91.0 PDMS on NAVSIM (vs. 89.6 for the Two-DiT variant), and demonstrates strong zero-shot transfer to nuScenes and Bench2Drive without fine-tuning, while ablations confirm the importance of single-DiT unification, video co-training, and mask-based modality decoupling.

cs.CV↗

arXiv2026

Towards Interactive Video World Modeling: Frontiers, Challenges, Benchmarks, and Future Trends

With rapid development of large language models and diffusion-based content generation, world modeling has attracted increasing research attention, benefiting various downstream domains such as game engines, embodied AI, autonomous driving, etc. Through explicitly incorporating user actions into world state transition, recent literature empowers world modeling with interactivity in an action-conditioned video or 3D generation paradigm, further enhancing controllability over world evolutions and facilitating users to freely traverse, manipulate, navigate, and personalize the state evolution. In this paper, we aim to systematically review recent research trends, technical developments, evaluation benchmarks, and also propose future potential directions in interactive world modeling. Specifically, we first summarize recent efforts and trends in terms of application scenarios, world state evolution, and scene modality. Afterwards, we delve into three crucial technical challenges, including action-conditioned controllability, long-horizon interactions and memory, and action-following responsiveness for real-time interactivity. Furthermore, we also thoroughly compare existing benchmarks and metrics in four specific application fields: open-world exploration, game engine, autonomous driving, and robotics. Finally, we discuss several promising future directions in achieving next-generation interactive world modeling. The corresponding repository is publicly available at: https://github.com/liujiuming123/Awesome-Interactive-World-Model.

cs.CV↗

arXiv2026

DriveVA: Video Action Models are Zero-Shot Drivers

Generalization is a central challenge in autonomous driving, as real-world deployment requires robust performance under unseen scenarios, sensor domains, and environmental conditions. Recent world-model-based planning methods have shown strong capabilities in scene understanding and multi-modal future prediction, yet their generalization across datasets and sensor configurations remains limited. In addition, their loosely coupled planning paradigm often leads to poor video-trajectory consistency during visual imagination. To overcome these limitations, we propose DriveVA, a novel autonomous driving world model that jointly decodes future visual forecasts and action sequences in a shared latent generative process. DriveVA inherits rich priors on motion dynamics and physical plausibility from well-pretrained large-scale video generation models to capture continuous spatiotemporal evolution and causal interaction patterns. To this end, DriveVA employs a DiT-based decoder to jointly predict future action sequences (trajectories) and videos, enabling tighter alignment between planning and scene evolution. We also introduce a video continuation strategy to strengthen long-duration rollout consistency. DriveVA achieves an impressive PDM-based planning performance of 90.9 PDM score on the NAVSIM benchmark. Extensive experiments also demonstrate the zero-shot capability and cross-domain generalization of DriveVA, which reduces average L2 error and collision rate by 78.9% and 83.3% on nuScenes and 52.5% and 52.4% on the Bench2Drive built on CARLA v2 compared with the state-of-the-art world-model-based planner.

cs.CV↗

arXiv2026

RegFormer++: An Efficient Large-Scale 3D LiDAR Point Registration Network with Projection-Aware 2D Transformer

Although point cloud registration has achieved remarkable advances in object-level and indoor scenes, large-scale LiDAR registration methods has been rarely explored before. Challenges mainly arise from the huge point scale, complex point distribution, and numerous outliers within outdoor LiDAR scans. In addition, most existing registration works generally adopt a two-stage paradigm: They first find correspondences by extracting discriminative local descriptors and then leverage robust estimators (e.g. RANSAC) to filter outliers, which are highly dependent on well-designed descriptors and post-processing choices. To address these problems, we propose a novel end-to-end differential transformer network, termed RegFormer++, for large-scale point cloud alignment without requiring any further post-processing. Specifically, a hierarchical projection-aware 2D transformer with linear complexity is proposed to project raw LiDAR points onto a cylindrical surface and extract global point features, which can improve resilience to outliers due to long-range dependencies. Because we fill original 3D coordinates into 2D projected positions, our designed transformer can benefit from both high efficiency in 2D processing and accuracy from 3D geometric information. Furthermore, to effectively reduce wrong point matching, a Bijective Association Transformer (BAT) is designed, combining both cross attention and all-to-all point gathering. To improve training stability and robustness, a feature-transformed optimal transport module is also designed for regressing the final pose transformation. Extensive experiments on KITTI, NuScenes, and Argoverse datasets demonstrate that our model achieves state-of-the-art performance in terms of both accuracy and efficiency.

cs.CV↗

arXiv2025

4DSTR: Advancing Generative 4D Gaussians with Spatial-Temporal Rectification for High-Quality and Consistent 4D Generation

Remarkable advances in recent 2D image and 3D shape generation have induced a significant focus on dynamic 4D content generation. However, previous 4D generation methods commonly struggle to maintain spatial-temporal consistency and adapt poorly to rapid temporal variations, due to the lack of effective spatial-temporal modeling. To address these problems, we propose a novel 4D generation network called 4DSTR, which modulates generative 4D Gaussian Splatting with spatial-temporal rectification. Specifically, temporal correlation across generated 4D sequences is designed to rectify deformable scales and rotations and guarantee temporal consistency. Furthermore, an adaptive spatial densification and pruning strategy is proposed to address significant temporal variations by dynamically adding or deleting Gaussian points with the awareness of their pre-frame movements. Extensive experiments demonstrate that our 4DSTR achieves state-of-the-art performance in video-to-4D generation, excelling in reconstruction quality, spatial-temporal consistency, and adaptation to rapid temporal movements.

cs.CV↗

arXiv2025

Variable-order fractional wave equation: Analysis, numerical approximation, and fast algorithm

We investigate a local modification of a variable-order fractional wave equation, which describes the propagation of diffusive wave in viscoelastic media with evolving physical property. We incorporate an equivalent formulation to prove the well-posedness of the model as well as its high order regularity estimates. To accommodate the convolution term in the reformulated model, we adopt the Ritz-Volterra finite element projection and then derive the rigorous error estimate for the fully-discretized finite element scheme. To circumvent the high computational cost from the temporal integral term, we exploit the translational invariance of the discrete coefficients associated with the convolution structure and construct a fast divide-and-conquer algorithm which reduces the computational complexity from $O(MN^2)$ to $O(MN\log^2 N)$. Numerical experiments are provided to verify the theoretical results and to demonstrate the accuracy and efficiency of the proposed method.

math.NA↗

arXiv2025

Multiscale solution decomposition of nonlocal-in-time problems with application in numerical computation

This work develops a multiscale solution decomposition (MSD) method for nonlocal-in-time problems to separate a series of known terms with multiscale singularity from the original singular solution such that the remaining unknown part becomes smoother. We demonstrate that the MSD provides a scenario where the smoothness assumption for solutions of weakly singular nonlocal-in-time problems, a commonly encountered assumption in numerous literature of numerical methods that is in general not true for original solutions, becomes appropriate such that abundant numerical analysis results therein become applicable. From computational aspect, instead of handling solution singularity, the MSD significantly reduces the numerical difficulties by separating and thus circumventing the solution singularity. We consider typical problems, including the fractional relaxation equation, Volterra integral equation, subdiffusion, integrodifferential equation and diffusion-wave equation, to demonstrate the universality of MSD and its effectiveness in improving the numerical accuracy or stability in comparison with classical methods.

math.NA↗

arXiv2025

DVLO4D: Deep Visual-Lidar Odometry with Sparse Spatial-temporal Fusion

Visual-LiDAR odometry is a critical component for autonomous system localization, yet achieving high accuracy and strong robustness remains a challenge. Traditional approaches commonly struggle with sensor misalignment, fail to fully leverage temporal information, and require extensive manual tuning to handle diverse sensor configurations. To address these problems, we introduce DVLO4D, a novel visual-LiDAR odometry framework that leverages sparse spatial-temporal fusion to enhance accuracy and robustness. Our approach proposes three key innovations: (1) Sparse Query Fusion, which utilizes sparse LiDAR queries for effective multi-modal data fusion; (2) a Temporal Interaction and Update module that integrates temporally-predicted positions with current frame data, providing better initialization values for pose estimation and enhancing model's robustness against accumulative errors; and (3) a Temporal Clip Training strategy combined with a Collective Average Loss mechanism that aggregates losses across multiple frames, enabling global optimization and reducing the scale drift over long sequences. Extensive experiments on the KITTI and Argoverse Odometry dataset demonstrate the superiority of our proposed DVLO4D, which achieves state-of-the-art performance in terms of both pose accuracy and robustness. Additionally, our method has high efficiency, with an inference time of 82 ms, possessing the potential for the real-time deployment.

cs.CV↗

arXiv2025

TopoLiDM: Topology-Aware LiDAR Diffusion Models for Interpretable and Realistic LiDAR Point Cloud Generation

LiDAR scene generation is critical for mitigating real-world LiDAR data collection costs and enhancing the robustness of downstream perception tasks in autonomous driving. However, existing methods commonly struggle to capture geometric realism and global topological consistency. Recent LiDAR Diffusion Models (LiDMs) predominantly embed LiDAR points into the latent space for improved generation efficiency, which limits their interpretable ability to model detailed geometric structures and preserve global topological consistency. To address these challenges, we propose TopoLiDM, a novel framework that integrates graph neural networks (GNNs) with diffusion models under topological regularization for high-fidelity LiDAR generation. Our approach first trains a topological-preserving VAE to extract latent graph representations by graph construction and multiple graph convolutional layers. Then we freeze the VAE and generate novel latent topological graphs through the latent diffusion models. We also introduce 0-dimensional persistent homology (PH) constraints, ensuring the generated LiDAR scenes adhere to real-world global topological structures. Extensive experiments on the KITTI-360 dataset demonstrate TopoLiDM's superiority over state-of-the-art methods, achieving improvements of 22.6% lower Frechet Range Image Distance (FRID) and 9.2% lower Minimum Matching Distance (MMD). Notably, our model also enables fast generation speed with an average inference time of 1.68 samples/s, showcasing its scalability for real-world applications. We will release the related codes at https://github.com/IRMVLab/TopoLiDM.

cs.CV↗

arXiv2025

Optimal control of variable-exponent subdiffusion

This work investigates the optimal control of the variable-exponent subdiffusion, which extends the work [Gunzburger and Wang, SIAM J. Control Optim. 2019] to the variable-exponent case to account for the multiscale and crossover diffusion behavior. To resolve the difficulties caused by the leading variable-exponent operator, we adopt the convolution method to reformulate the model into an equivalent but more tractable form, and then prove the well-posedness and weighted regularity of the optimal control. As the convolution kernels in reformulated models are indefinite-sign, non-positive-definite, and non-monotonic, we adopt the discrete convolution kernel approach in numerical analysis to show the $O(\tau(1+|\ln\tau|)+h^2)$ accuracy of the schemes for state and adjoint equations. Numerical experiments are performed to substantiate the theoretical findings.

math.OC↗

arXiv2025

Development_of_a_novel_high-performance_balanced_homodyne_detector

True random numbers are extracted through measurements of vacuum fluctuations in quantum state components. We propose an improved scheme utilizing an optimization-based simulation methodology to enhance the temporal resolution of quantum state detection and processing efficiency of vacuum fluctuation signals in continuous-variable quantum random number generators (CV-QRNGs), while simultaneously maximizing the entropy content of quantum noise sources. This work presents the first application of optimization simulation methodology to balanced homodyne detector (BHD) circuit design, with particular emphasis on improving high-frequency transmission characteristics. The design framework prioritizes system stability and S-parameter sensitivity to optimize both circuit architecture and critical component parameters. The AC amplifier circuit was implemented through ADS high-frequency simulations using two ABA-52563 RF amplifiers in a cascaded configuration, with circuit modeling performed on Rogers 4350 substrate optimized for high-frequency applications. This approach enabled the development of a switched-configuration BHD featuring: 1) 1.9 GHz bandwidth, 2) 41.5 dB signal-to-noise ratio at 1.75 GHz, 3) 30 dB common-mode rejection ratio at 100 MHz, and 4) frequency response flatness within 1.5 dB across 1.3-1.7 GHz. Additionally, the Husimi function is employed for entropy analysis to reconstruct vacuum state phase-space distributions, validating the detector's quantum measurement fidelity. The implemented system demonstrates a collective generation rate of 20.0504 Gbps across four parallel channels, with all output streams successfully passing NIST SP 800-22 statistical testing requirements.

quant-ph↗

arXiv2024

Generalizing subdiffusive Black-Scholes model by variable exponent: Model transformation and numerical approximation

This work generalizes the subdiffusive Black-Scholes model by introducing the variable exponent in order to provide adequate descriptions for the option pricing, where the variable exponent may account for the variation of the memory property. In addition to standard nonlinear-to-linear transformation, we apply a further spatial-temporal transformation to convert the model to a more tractable form in order to circumvent the difficulties caused by the ``non-positive, non-monotonic'' variable-exponent memory kernel. An interesting phenomenon is that the spatial transformation not only eliminates the advection term but naturally turns the original noncoercive spatial operator into a coercive one due to the specific structure of the Black-Scholes model, which thus avoids imposing constraints on coefficients. Then we perform numerical analysis for both the semi-discrete and fully discrete schemes to support numerical simulation. Numerical experiments are carried out to substantiate the theoretical results.

math.NA↗

arXiv2024

Factual Serialization Enhancement: A Key Innovation for Chest X-ray Report Generation

A radiology report comprises presentation-style vocabulary, which ensures clarity and organization, and factual vocabulary, which provides accurate and objective descriptions based on observable findings. While manually writing these reports is time-consuming and labor-intensive, automatic report generation offers a promising alternative. A critical step in this process is to align radiographs with their corresponding reports. However, existing methods often rely on complete reports for alignment, overlooking the impact of presentation-style vocabulary. To address this issue, we propose FSE, a two-stage Factual Serialization Enhancement method. In Stage 1, we introduce factuality-guided contrastive learning for visual representation by maximizing the semantic correspondence between radiographs and corresponding factual descriptions. In Stage 2, we present evidence-driven report generation that enhances diagnostic accuracy by integrating insights from similar historical cases structured as factual serialization. Experiments on MIMIC-CXR and IU X-ray datasets across specific and general scenarios demonstrate that FSE outperforms state-of-the-art approaches in both natural language generation and clinical efficacy metrics. Ablation studies further emphasize the positive effects of factual serialization in Stage 1 and Stage 2. The code is available at https://github.com/mk-runner/FSE.

eess.IV↗

arXiv2023

Road Network Guided Fine-Grained Urban Traffic Flow Inference

Accurate inference of fine-grained traffic flow from coarse-grained one is an emerging yet crucial problem, which can help greatly reduce the number of the required traffic monitoring sensors for cost savings. In this work, we notice that traffic flow has a high correlation with road network, which was either completely ignored or simply treated as an external factor in previous works. To facilitate this problem, we propose a novel Road-Aware Traffic Flow Magnifier (RATFM) that explicitly exploits the prior knowledge of road networks to fully learn the road-aware spatial distribution of fine-grained traffic flow. Specifically, a multi-directional 1D convolutional layer is first introduced to extract the semantic feature of the road network. Subsequently, we incorporate the road network feature and coarse-grained flow feature to regularize the short-range spatial distribution modeling of road-relative traffic flow. Furthermore, we take the road network feature as a query to capture the long-range spatial distribution of traffic flow with a transformer architecture. Benefiting from the road-aware inference mechanism, our method can generate high-quality fine-grained traffic flow maps. Extensive experiments on three real-world datasets show that the proposed RATFM outperforms state-of-the-art models under various scenarios. Our code and datasets are released at {\url{https://github.com/luimoli/RATFM}}.

cs.LG↗

arXiv2023

Tracing the Influence of Predecessors on Trajectory Prediction

In real-world traffic scenarios, agents such as pedestrians and car drivers often observe neighboring agents who exhibit similar behavior as examples and then mimic their actions to some extent in their own behavior. This information can serve as prior knowledge for trajectory prediction, which is unfortunately largely overlooked in current trajectory prediction models. This paper introduces a novel Predecessor-and-Successor (PnS) method that incorporates a predecessor tracing module to model the influence of predecessors (identified from concurrent neighboring agents) on the successor (target agent) within the same scene. The method utilizes the moving patterns of these predecessors to guide the predictor in trajectory prediction. PnS effectively aligns the motion encodings of the successor with multiple potential predecessors in a probabilistic manner, facilitating the decoding process. We demonstrate the effectiveness of PnS by integrating it into a graph-based predictor for pedestrian trajectory prediction on the ETH/UCY datasets, resulting in a new state-of-the-art performance. Furthermore, we replace the HD map-based scene-context module with our PnS method in a transformer-based predictor for vehicle trajectory prediction on the nuScenes dataset, showing that the predictor maintains good prediction performance even without relying on any map information.

cs.RO↗

arXiv2023

An End-to-End Framework of Road User Detection, Tracking, and Prediction from Monocular Images

Perception that involves multi-object detection and tracking, and trajectory prediction are two major tasks of autonomous driving. However, they are currently mostly studied separately, which results in most trajectory prediction modules being developed based on ground truth trajectories without taking into account that trajectories extracted from the detection and tracking modules in real-world scenarios are noisy. These noisy trajectories can have a significant impact on the performance of the trajectory predictor and can lead to serious prediction errors. In this paper, we build an end-to-end framework for detection, tracking, and trajectory prediction called ODTP (Online Detection, Tracking and Prediction). It adopts the state-of-the-art online multi-object tracking model, QD-3DT, for perception and trains the trajectory predictor, DCENet++, directly based on the detection results without purely relying on ground truth trajectories. We evaluate the performance of ODTP on the widely used nuScenes dataset for autonomous driving. Extensive experiments show that ODPT achieves high performance end-to-end trajectory prediction. DCENet++, with the enhanced dynamic maps, predicts more accurate trajectories than its base model. It is also more robust when compared with other generative and deterministic trajectory prediction models trained on noisy detection results.

cs.CV↗

arXiv2023

Reduced-Complexity Cross-Domain Iterative Detection for OTFS Modulation via Delay-Doppler Decoupling

In this paper, a reduced-complexity cross-domain iterative detection for orthogonal time frequency space (OTFS) modulation is proposed, which exploits channel properties in both time and delay-Doppler domains. Specifically, we first show that in the time domain effective channel, the path delay only introduces interference among samples in adjacent time slots, while the Doppler becomes a phase term that does not affect the channel sparsity. This ``band-limited'' matrix structure motivates us to apply a reduced-size linear minimum mean square error (LMMSE) filter to eliminate the effect of delay in the time domain, while exploiting the cross-domain iteration for minimizing the effect of Doppler by noticing that the time and Doppler are a pair of Fourier dual. The state (MSE) evolution was derived and compared with bounds to verify the effectiveness of the proposed scheme. Simulation results demonstrate that the proposed scheme achieves almost the same error performance as the optimal detection, but only requires a reduced complexity.

cs.IT↗