arXiv subjects

Juan Zhu

Publications and source records attributed to Juan Zhu.

7 recordsLinked to original sources

arXiv2026

AT-VLA: Adaptive Tactile Injection for Enhanced Feedback Reaction in Vision-Language-Action Models

Vision-Language-Action (VLA) models have significantly advanced the capabilities of robotic agents in executing diverse tasks; however, they still face challenges in contact-rich manipulation scenarios that require precise physical interactions. To address this limitation, recent studies have attempted to incorporate tactile signals during downstream tasks, enabling pretrained VLAs to interpret tactile feedback. Nevertheless, introducing new modalities during finetuning, which are rarely present in the pretrain stage, may disrupt the pretrained capabilities of VLAs. In addition, the inherently slow inference speed of VLAs hampers real-time responsiveness and limits the effective utilization of tactile feedback for action adjustment. To overcome these challenges, we propose Adaptive Tactile Vision-Language-Action (AT-VLA), which introduces a novel Adaptive Tactile Injection mechanism. This mechanism dynamically determines the appropriate timing and locations for tactile injection, incorporating only when it significantly contributes to action generation, thereby minimizing interference with pretrained representations. Furthermore, to enable rapid and accurate tactile responses, we propose a Tactile Reaction Dual-Stream mechanism, which decouples sensory processing into a slow visual-language stream for low-frequency perceptual reasoning and a fast tactile control stream for high-frequency physical interaction understanding, achieving real-time close-loop responses within 0.04 s. Real-world experiments thoroughly validate the effectiveness of AT-VLA in contact-rich manipulation tasks. The project page is available at: https://sites.google.com/view/at-vla.

cs.RO↗

arXiv2026

AnchorVLA4D: an Anchor-Based Spatial-Temporal Vision-Language-Action Model for Robotic Manipulation

Since current Vision-Language-Action (VLA) systems suffer from limited spatial perception and the absence of memory throughout manipulation, we investigate visual anchors as a means to enhance spatial and temporal reasoning within VLA policies for robotic manipulation. Conventional VLAs generate actions by conditioning on a single current frame together with a language instruction. However, since the frame is encoded as a 2D image, it does not contain detailed spatial information, and the VLA similarly lacks any means to incorporate past context. As a result, it frequently forgets objects under occlusion and becomes spatially disoriented during the manipulation process. Thus, we propose AnchorVLA4D, a simple spatial-temporal VLA that augments the visual input with an anchor image to preserve the initial scene context throughout execution, and adds a lightweight spatial encoder that jointly processes the anchor and current frames to expose geometric relationships within an episode. Built on a Qwen2.5-VL backbone with a diffusion-based action head, AnchorVLA4D requires no additional sensing modalities (e.g., depth or point clouds) and introduces negligible inference overhead. Combining anchoring with a frozen pretrained spatial encoder yields further gains, realizing a 13.6% improvement on the Simpler WidowX benchmark and confirming the approach on real-world tasks, where it achieved an average success rate of 80%.

cs.RO↗

arXiv2026

Modular Foundation Model Inference at the Edge: Network-Aware Microservice Optimization

Foundation models (FMs) unlock unprecedented multimodal and multitask intelligence, yet their cloud-centric deployment precludes real-time responsiveness and compromises user privacy. Meanwhile, monolithic execution at the edge remains infeasible under stringent resource limits and uncertain network dynamics. To bridge this gap, we propose a microservice-based FM inference framework that exploits the intrinsic functional asymmetry between heavyweight core services and agile light services. Our two-tier deployment strategy ensures robust Quality of Service (QoS) under resource contention. Specifically, core services are placed statically via a long-term network-aware integer program with sparsity constraints to form a fault-tolerant backbone. On the other hand, light services are orchestrated dynamically by a low-complexity online controller that integrates effective capacity theory with Lyapunov optimization, providing probabilistic latency guarantees under real-time workload fluctuations. Simulations demonstrate that our framework achieves over 84% average on-time task completion with moderate deployment costs and maintains strong robustness as the system load scales.

cs.DC↗

arXiv2025

Synergistic Tensor and Pipeline Parallelism

In the machine learning system, the hybrid model parallelism combining tensor parallelism (TP) and pipeline parallelism (PP) has become the dominant solution for distributed training of Large Language Models~(LLMs) and Multimodal LLMs (MLLMs). However, TP introduces significant collective communication overheads, while PP suffers from synchronization inefficiencies such as pipeline bubbles. Existing works primarily address these challenges from isolated perspectives, focusing either on overlapping TP communication or on flexible PP scheduling to mitigate pipeline bubbles. In this paper, we propose a new synergistic tensor and pipeline parallelism schedule that simultaneously reduces both types of bubbles. Our proposed schedule decouples the forward and backward passes in PP into fine-grained computation units, which are then braided to form a composite computation sequence. This compositional structure enables near-complete elimination of TP-related bubbles. Building upon this structure, we further design the PP schedule to minimize PP bubbles. Experimental results demonstrate that our approach improves training throughput by up to 12% for LLMs and 16% for MLLMs compared to existing scheduling methods. Our source code is avaiable at https://github.com/MICLAB-BUPT/STP.

cs.DC↗

arXiv2025

Imagine2Act: Leveraging Object-Action Motion Consistency from Imagined Goals for Robotic Manipulation

Relational object rearrangement (ROR) tasks (e.g., insert flower to vase) require a robot to manipulate objects with precise semantic and geometric reasoning. Existing approaches either rely on pre-collected demonstrations that struggle to capture complex geometric constraints or generate goal-state observations to capture semantic and geometric knowledge, but fail to explicitly couple object transformation with action prediction, resulting in errors due to generative noise. To address these limitations, we propose Imagine2Act, a 3D imitation-learning framework that incorporates semantic and geometric constraints of objects into policy learning to tackle high-precision manipulation tasks. We first generate imagined goal images conditioned on language instructions and reconstruct corresponding 3D point clouds to provide robust semantic and geometric priors. These imagined goal point clouds serve as additional inputs to the policy model, while an object-action consistency strategy with soft pose supervision explicitly aligns predicted end-effector motion with generated object transformation. This design enables Imagine2Act to reason about semantic and geometric relationships between objects and predict accurate actions across diverse tasks. Experiments in both simulation and the real world demonstrate that Imagine2Act outperforms previous state-of-the-art policies. More visualizations can be found at https://sites.google.com/view/imagine2act.

cs.RO↗

arXiv2025

Spindle: Efficient Distributed Training of Multi-Task Large Models via Wavefront Scheduling

Recent foundation models are capable of handling multiple tasks and multiple data modalities with the unified base model structure and several specialized model components. However, efficient training of such multi-task (MT) multi-modal (MM) models poses significant system challenges due to the sophisticated model architecture and the heterogeneous workloads of different tasks and modalities. In this paper, we propose Spindle, a brand new training system tailored for resource-efficient and high-performance training of MT MM models via wavefront scheduling. The key idea of Spindle is to decompose the model execution into waves and address the joint optimization problem sequentially, including both heterogeneity-aware workload parallelization and dependency-driven execution scheduling. We build our system and evaluate it on various MT MM models. Experiments demonstrate the superior performance and efficiency of Spindle, with speedup ratio up to 71% compared to state-of-the-art training systems.

cs.DC↗

arXiv2016

Dip-coating with prestructured substrates: transfer of simple liquids and Langmuir-Blodgett monolayers

When a plate is withdrawn from a liquid bath, either a static meniscus forms in the transition region between the bath and the substrate or a liquid film of finite thickness (a Landau-Levich film) is transferred onto the moving substrate. If the substrate is inhomogeneous, e.g., has a prestructure consisting of stripes of different wettabilities, the meniscus can be deformed or show a complex dynamic behavior. Here we study the free surface shape and dynamics of a dragged meniscus occurring for striped prestructures with two orientations, parallel and perpendicular to the transfer direction. A thin film model is employed that accounts for capillarity through a Laplace pressure and for the spatially varying wettability through a Derjaguin (or disjoining) pressure. Numerical continuation is used to obtain steady free surface profiles and corresponding bifurcation diagrams in the case of substrates with different homogeneous wettabilities. Direct numerical simulations are employed in the case of the various striped prestructures. The final part illustrates the importance of our findings for particular applications that involve complex liquids by modeling a Langmuir-Blodgett transfer experiment. There, one transfers a monolayer of an insoluble surfactant that covers the surface of the bath onto the moving substrate. The resulting pattern formation phenomena can be crucially influenced by the hydrodynamics of the liquid meniscus that itself depends on the prestructure on the substrate. In particular, we show how prestructure stripes parallel to the transfer direction lead to the formation of bent stripes in the surfactant coverage after transfer and present similar experimental results.

physics.flu-dyn↗