arXiv subjects

Mingjie Wei

Publications and source records attributed to Mingjie Wei.

10 recordsLinked to original sources

arXiv2026

RLinf-VLA: A Unified and Efficient Framework for Reinforcement Learning of Vision-Language-Action Models

Recent studies have demonstrated the potential of reinforcement learning (RL) to improve the task performance of vision-language-action (VLA) models through interaction. However, current efforts remain fragmented, lacking a unified platform for fair comparison across architectures and algorithms, as well as an efficient system design for scalable training. Therefore, we present RLinf-VLA, a unified and efficient framework for scalable RL training of VLA models. RLinf-VLA standardizes the integration of diverse VLA architectures, RL algorithms, and heterogeneous simulators through a unified interface, enabling extensibility and reproducibility. To improve efficiency, the framework adopts a flexible resource allocation architecture for rendering, inference, and training in RL pipelines. In particular, RLinf-VLA introduces a hybrid fine-grained pipeline allocation strategy that achieves a 1.61$\times$-1.88$\times$ training speedup on ManiSkill. Using this framework, RL-trained models achieve strong performance across embodied benchmarks, including 98.11% success on 130 LIBERO tasks, 97.66% success on 25 ManiSkill tasks, and 84.63% average success across 6 RoboTwin tasks. In addition, RLinf-VLA distills a set of effective practices for RL-based VLA training. We envision RLinf-VLA as a foundational framework for efficient, unified, and reproducible research in embodied intelligence.

cs.RO↗

arXiv2026

WoVR: World Models as Reliable Simulators for Post-Training VLA Policies with RL

Reinforcement learning (RL) promises to unlock capabilities beyond imitation learning for Vision--Language--Action (VLA) models, but its requirement for massive real-world interaction prevents direct deployment on physical robots. Recent work attempts to use learned world models as simulators for policy optimization, yet closed-loop imagined rollouts inevitably suffer from hallucination and long-horizon error accumulation. Such errors not only degrade visual fidelity, but also mislead policy optimization by providing unreliable learning signals. We propose WoVR, a reliable world-model-based RL framework for post-training VLA policies. Instead of assuming a faithful world model, WoVR explicitly regulates how RL interacts with imperfect imagined dynamics. It improves rollout stability through a controllable action-conditioned video world model, reshapes imagined interaction to reduce effective error depth via Keyframe-Initialized Rollouts, and maintains policy--simulator alignment through World Model-Policy co-evolution. Extensive experiments demonstrate that WoVR enables stable long-horizon imagined rollouts and effective policy optimization, achieving superior LIBERO performance and consistent real-world gains across multiple robotic platforms. These results show that world models can serve as practical simulators for RL when hallucination is explicitly controlled. Additional visualization results are available at https://wovr-corl.github.io.

cs.RO↗

arXiv2026

Tex3D: Objects as Attack Surfaces via Adversarial 3D Textures for Vision-Language-Action Models

Vision-language-action (VLA) models have shown strong performance in robotic manipulation, yet their robustness to physically realizable adversarial attacks remains underexplored. Existing studies reveal vulnerabilities through language perturbations and 2D visual attacks, but these attack surfaces are either less representative of real deployment or limited in physical realism. In contrast, adversarial 3D textures pose a more physically plausible and damaging threat, as they are naturally attached to manipulated objects and are easier to deploy in physical environments. Bringing adversarial 3D textures to VLA systems is nevertheless nontrivial. A central obstacle is that standard 3D simulators do not provide a differentiable optimization path from the VLA objective function back to object appearance, making it difficult to optimize through an end-to-end manner. To address this, we introduce Foreground-Background Decoupling (FBD), which enables differentiable texture optimization through dual-renderer alignment while preserving the original simulation environment. To further ensure that the attack remains effective across long-horizon and diverse viewpoints in the physical world, we propose Trajectory-Aware Adversarial Optimization (TAAO), which prioritizes behaviorally critical frames and stabilizes optimization with a vertex-based parameterization. Built on these designs, we present Tex3D, the first framework for end-to-end optimization of 3D adversarial textures directly within the VLA simulation environment. Experiments in both simulation and real-robot settings show that Tex3D significantly degrades VLA performance across multiple manipulation tasks, achieving task failure rates of up to 96.7\%. Our empirical results expose critical vulnerabilities of VLA systems to physically grounded 3D adversarial attacks and highlight the need for robustness-aware training.

cs.CV↗

arXiv2025

RLinf: Flexible and Efficient Large-scale Reinforcement Learning via Macro-to-Micro Flow Transformation

Reinforcement learning (RL) has demonstrated immense potential in advancing artificial general intelligence, agentic intelligence, and embodied intelligence. However, the inherent heterogeneity and dynamicity of RL workflows often lead to low hardware utilization and slow training on existing systems. In this paper, we present RLinf, a high-performance RL training system based on our key observation that the major roadblock to efficient RL training lies in system flexibility. To maximize flexibility and efficiency, RLinf is built atop a novel RL system design paradigm called macro-to-micro flow transformation (M2Flow), which automatically breaks down high-level, easy-to-compose RL workflows at both the temporal and spatial dimensions, and recomposes them into optimized execution flows. Supported by RLinf worker's adaptive communication capability, we devise context switching and elastic pipelining to realize M2Flow transformation, and a profiling-guided scheduling policy to generate optimal execution plans. Extensive evaluations on both reasoning RL and embodied RL tasks demonstrate that RLinf consistently outperforms state-of-the-art systems, achieving $1.07\times-2.43\times$ speedup in end-to-end training throughput.

cs.LG↗

arXiv2025

LLaPa: A Vision-Language Model Framework for Counterfactual-Aware Procedural Planning

While large language models (LLMs) have advanced procedural planning for embodied AI systems through strong reasoning abilities, the integration of multimodal inputs and counterfactual reasoning remains underexplored. To tackle these challenges, we introduce LLaPa, a vision-language model framework designed for multimodal procedural planning. LLaPa generates executable action sequences from textual task descriptions and visual environmental images using vision-language models (VLMs). Furthermore, we enhance LLaPa with two auxiliary modules to improve procedural planning. The first module, the Task-Environment Reranker (TER), leverages task-oriented segmentation to create a task-sensitive feature space, aligning textual descriptions with visual environments and emphasizing critical regions for procedural execution. The second module, the Counterfactual Activities Retriever (CAR), identifies and emphasizes potential counterfactual conditions, enhancing the model's reasoning capability in counterfactual scenarios. Extensive experiments on ActPlan-1K and ALFRED benchmarks demonstrate that LLaPa generates higher-quality plans with superior LCS and correctness, outperforming advanced models. The code and models are available https://github.com/sunshibo1234/LLaPa.

cs.CL↗

arXiv2025

Learning Pyramid-structured Long-range Dependencies for 3D Human Pose Estimation

Action coordination in human structure is indispensable for the spatial constraints of 2D joints to recover 3D pose. Usually, action coordination is represented as a long-range dependence among body parts. However, there are two main challenges in modeling long-range dependencies. First, joints should not only be constrained by other individual joints but also be modulated by the body parts. Second, existing methods make networks deeper to learn dependencies between non-linked parts. They introduce uncorrelated noise and increase the model size. In this paper, we utilize a pyramid structure to better learn potential long-range dependencies. It can capture the correlation across joints and groups, which complements the context of the human sub-structure. In an effective cross-scale way, it captures the pyramid-structured long-range dependence. Specifically, we propose a novel Pyramid Graph Attention (PGA) module to capture long-range cross-scale dependencies. It concatenates information from various scales into a compact sequence, and then computes the correlation between scales in parallel. Combining PGA with graph convolution modules, we develop a Pyramid Graph Transformer (PGFormer) for 3D human pose estimation, which is a lightweight multi-scale transformer architecture. It encapsulates human sub-structures into self-attention by pooling. Extensive experiments show that our approach achieves lower error and smaller model size than state-of-the-art methods on Human3.6M and MPI-INF-3DHP datasets. The code is available at https://github.com/MingjieWe/PGFormer.

cs.CV↗

arXiv2025

ACMo: Attribute Controllable Motion Generation

Attributes such as style, fine-grained text, and trajectory are specific conditions for describing motion. However, existing methods often lack precise user control over motion attributes and suffer from limited generalizability to unseen motions. This work introduces an Attribute Controllable Motion generation architecture, to address these challenges via decouple any conditions and control them separately. Firstly, we explored the Attribute Diffusion Model to imporve text-to-motion performance via decouple text and motion learning, as the controllable model relies heavily on the pre-trained model. Then, we introduce Motion Adpater to quickly finetune previously unseen motion patterns. Its motion prompts inputs achieve multimodal text-to-motion generation that captures user-specified styles. Finally, we propose a LLM Planner to bridge the gap between unseen attributes and dataset-specific texts via local knowledage for user-friendly interaction. Our approach introduces the capability for motion prompts for stylize generation, enabling fine-grained and user-friendly attribute control while providing performance comparable to state-of-the-art methods. Project page: https://mjwei3d.github.io/ACMo/

cs.CV↗

arXiv2021

Nonlinearity Characteristic of High Impedance Fault at Resonant Distribution Networks: Theoretical Basis to Identify the Faulty Feeder

Feeder identification is indispensable for distribution networks to locate faults at a specific feeder, especially when measuring de-vices are insufficient for precise locations. For the high imped-ance fault (HIF), the feeder identification is much more compli-cated and related approaches are still in the early stage. This paper thoroughly and theoretically reveals the features of dif-ferent feeders when a HIF happens at the resonant grounded neutral (RGN) network, which is the most challenging condition for feeder identification. Firstly, the diversity of nonlinearity existing in HIFs is explained from the aspect of energy. Then, the differences of nonlinearities of zero-sequence currents between healthy and faulty feeders are deduced theoretically. Variations of the detuning index and damping ratio that exist in industries are both considered. Afterward, these theoretical conclusions are verified by the HIF cases experimented in a real 10kV system. Finally, based on the theories, we discuss about why the existing approaches are not reliable enough, and suggest some improve-ments.

eess.SY↗

arXiv2020

Detection and Feeder Identification of the High Impedance Fault at Distribution Networks Based on Synchronous Waveform Distortions

Diagnosis of high impedance fault (HIF) is a challenge for nowadays distribution network protections. The fault current of a HIF is much lower than that of a normal load, and fault feature is significantly affected by fault scenarios. A detection and feeder identification algorithm for HIFs is proposed in this paper, based on the high-resolution and synchronous waveform data. In the algorithm, an interval slope is defined to describe the waveform distortions, which guarantees a uniform feature description under various HIF nonlinearities and noise interferences. For three typical types of network neutrals, i.e.,isolated neutral, resonant neutral, and low-resistor-earthed neutral, differences of the distorted components between the zero-sequence currents of healthy and faulty feeders are mathematically deduced, respectively. As a result, the proposed criterion, which is based on the distortion relationships between zero-sequence currents of feeders and the zero-sequence voltage at the substation, is theoretically supported. 28 HIFs grounded to various materials are tested in a 10kV distribution networkwith three neutral types, and are utilized to verify the effectiveness of the proposed algorithm.

eess.SY↗

arXiv2020

Distortion-Based Detection of High Impedance Fault in Distribution Systems

Detecting the High impedance fault (HIF) in distribution systems plays an important role in power utilization safety. However, many HIFs are challenging to be identified due to their low currents and diverse characteristics. In particular, the slight nonlinearity during weak arcing processes, the distortion offset caused by the lag of heat dissipations, and the interference of background noises could lead to invalid of traditional detection algorithms. This paper proposes a distortion-based algorithm to improve the reliability of HIF detection. Firstly, the challenges brought by the diversity of HIF characteristics are illustrated with the experiments in a 10kV distribution system. Then, HIFs are classified into five types according to their characteristics. Secondly, an interval slope is defined to describe the waveform distortions of HIFs, and is extracted with the methods of the linear least square filtering (LLSF) and the Grubbs-criterion-based robust local regression smoothing (Grubbs-RLRS), so that feature descriptions under different fault conditions can be unified. Thirdly, an algorithm, as well as the criteria, is proposed to identify the fault features described by the interval slopes. Finally, the detection reliability of the algorithm is thoroughly verified with field HIF data, and the results show the improvements by comparing it to other ad-vanced algorithms.

eess.SY↗