arXiv subjects

Donggang Cao

Publications and source records attributed to Donggang Cao.

8 recordsLinked to original sources

arXiv2026

DyQ-VLA: Temporal-Dynamic-Aware Quantization for Embodied Vision-Language-Action Models

Vision-language-action (VLA) models achieve favorable task performance, yet runtime errors in closed-loop execution evolve with alternating updates of actions and observations. Existing VLA quantization methods mainly trade off inference speed and model performance while rarely investigating how quantization alters runtime errors. By comparing error distributions between full-precision and quantized models, we identify distinct evolutionary patterns for the two types of errors: quantization enlarges the variance of translation error distributions and increases their dispersion, whereas the distribution center of rotation errors gradually shifts across execution steps, demonstrating cumulative drift. Motivated by this observation, we rethink the optimal quantization strategy for VLA models and propose \textit{DyQ-VLA}, a runtime-error-aware quantization framework, which dynamically selects activation precision according to execution steps and tracks as well as compensates rotation errors via accumulated quantization residuals. Corresponding operators and runtime adaptation strategies are devised within the framework to enable dynamic-precision execution. Experiments show that \textit{DyQ-VLA} achieves a 1.88 to 1.93 inference speedups while maintaining comparable or higher average task success rates. Moreover, it reduces mean execution steps by 17.7% to 19.8%. Our code is here: https://anonymous.4open.science/r/DyQ-VLA-7F51/.

cs.LG↗

arXiv2026

VLN-Cache: Enabling Token Caching for VLN Models with Visual/Semantic Dynamics Awareness

Vision-and-Language Navigation (VLN) increasingly relies on large vision-language models, but their inference cost conflicts with real-time deployment. Token caching is a promising training-free strategy that avoids redundant computation by reusing stable visual tokens across frames. However, existing methods assume a static camera and fixed semantic focus, assumptions that VLN fundamentally violates. We identify two failure modes: (1) visual dynamics, where viewpoint shift displaces token positions across frames, causing position-wise matching to pair misaligned content; (2) semantic dynamics, where token relevance shifts across task stages as navigation progresses, making cached states stale. We propose VLN-Cache, a visual-dynamic-aware and semantic-dynamic-aware caching framework that introduces view-aligned remapping to recover geometric correspondences and a task-relevance saliency filter to veto reuse at semantic transitions. A layer-adaptive entropy policy further balances the per-layer reuse budget. Experiments on the R2R-CE simulation benchmark show up to 1.52x speedup while maintaining competitive navigation success rates.

cs.RO↗

arXiv2026

KERV: Kinematic-Rectified Speculative Decoding for Embodied VLA Models

Vision-Language-Action (VLA) models build a token-domain robot control paradigm, yet suffer from low speed. Speculative Decoding (SD) is an optimization strategy that can boost inference speed. Two key issues emerge when integrating VLA and SD: first, SD relies on re-inference to address token errors, which is computationally expensive; second, to mitigate token errors, the acceptance threshold in SD requires careful adjustment. Existing works fail to address the above two issues effectively. Meanwhile, as the bridge between AI and the physical world, existing embodied intelligence has overlooked the application of robotic kinematics. To address these issues, we innovatively combine token-domain VLA models with kinematic-domain prediction for SD, proposing a kinematic-rectified SD framework named KERV. We employ a kinematics-based Kalman Filter to predict actions and compensate for SD errors, avoiding costly re-inference. Moreover, we design a kinematics-based adjustment strategy to dynamically rectify the acceptance threshold, addressing the difficulty of threshold determination. Experimental results across diverse tasks and environments demonstrate that KERV achieves 27%~37% acceleration with nearly no Success Rate loss.

cs.RO↗

arXiv2026

HeiSD: Hybrid Speculative Decoding for Embodied Vision-Language-Action Models with Kinematic Awareness

Vision-Language-Action (VLA) Models have become the mainstream solution for robot control, but suffer from slow inference speeds. Speculative Decoding (SD) is a promising acceleration method which can be divided into two categories: drafter-based SD and retrieval-based SD. Each of the two methods demonstrates complementary advantages and limitations when applied to VLA models, leading to the hypothesis that a hybrid approach integrating these two methods will yield better performance. In this paper, we first conduct a series of detailed analyses to reveal the advantages and feasibility of hybrid utilization. However, even with the aforementioned key insights, implementing hybrid SD in VLA models presents several challenges: (1) draft rejection and persistent errors in retrieval-based SD; (2) difficulty in determining the hybrid boundary. To address these, we propose the HeiSD framework. We propose a retrieval-based SD optimization method in HeiSD, which contains a verify-skip mechanism and a sequence-wise relaxed acceptance strategy. Moreover, we proposed a kinematic-based fused metric in HeiSD to automatically determine the hybrid boundary. Experimental results demonstrate that HeiSD attains a speedup of up to 2.45x in simulation benchmarks and 2.06x~2.41x in real-world scenarios, while sustaining a high task success rate.

cs.RO↗

arXiv2025

Optoelectronically Directed Self-Assembly of Active and Passive Particles into Programmable and Reconfigurable Colloidal Structures

Controlled assembly of active-passive colloidal mixtures offers a route to reconfigurable microscale machines, but their self-assembly pathways remain poorly understood. We study the directed assembly of metallo-dielectric Janus particles (JPs) and passive polystyrene (PS) beads using optoelectrically reconfigurable AC-field patterning, which allows precise control over particle composition and binding sequence. Through experiments, analytical modeling, and simulations, we show that dipolar interactions drive robust JP-JP and JP-PS dimer formation with frequency-dependent stability. At intermediate and high frequencies, JP-PS binding is strongly attractive, whereas at low frequencies it becomes effectively repulsive due to electrical double-layer screening and electrohydrodynamic flows at the metallic hemisphere. In multi-particle systems, PS beads act as cooperative hubs that hierarchically recruit JPs, yielding higher-order hybrid structures. We identify structural isomers - for example, 3JP + 1PS clusters can form chain-like or triangular configurations depending on assembly sequence. Simulations confirm both as equilibrium states, with the triangular isomer slightly more stable. Similar polymorphism appears in larger clusters (4JPs). Overall, we establish a framework for controlled active-passive colloidal assembly, showing how frequency-tunable interactions and structural polymorphism enable the design of reconfigurable colloidal machines for applications in microrobotics, targeted delivery, and adaptive materials.

cond-mat.soft↗

arXiv2025

Dielectrophoretic response and electro-deformation of soft bioparticles interacting with a metallo-dielectric Janus active particle

Active (self-propelling) particles have emerged as innovative microscale tools in the field of single cell analysis with the advantages of being untethered, remotely controlled, hybrid powered, with sub-cellular precision. This study investigates the dielectrophoretic (DEP) response and electro-mechanical deformation of cell nuclei interacting with active metallo-dielectric Janus Particles (JPs) under an externally applied electric field. An equivalent droplet two-phase model is employed to simulate the bioparticle, coupling the Navier-Stokes equations with the Phase Field Model to capture fluid motion and interface dynamics. Good qualitative agreement is obtained among experimental, analytical, and numerical results. The findings reveal a nonlinear relationship between nucleus deformation and its surface coverage of the JP with respect to the applied voltage. The overall coverage ratio of the JP dielectric hemisphere increases with voltage as the positive DEP force on the dielectric side strengthens, exhibiting maximum at a certain voltage. The strong correlation between nucleus flexibility and JP surface coverage suggests that the JP coverage ratio could serve as a biomechanical marker for nucleus deformability, providing a novel method for in-situ evaluation of nucleus mechanics.

cond-mat.soft↗

arXiv2020

SpotTune: Leveraging Transient Resources for Cost-efficient Hyper-parameter Tuning in the Public Cloud

Hyper-parameter tuning (HPT) is crucial for many machine learning (ML) algorithms. But due to the large searching space, HPT is usually time-consuming and resource-intensive. Nowadays, many researchers use public cloud resources to train machine learning models, convenient yet expensive. How to speed up the HPT process while at the same time reduce cost is very important for cloud ML users. In this paper, we propose SpotTune, an approach that exploits transient revocable resources in the public cloud with some tailored strategies to do HPT in a parallel and cost-efficient manner. Orchestrating the HPT process upon transient servers, SpotTune uses two main techniques, fine-grained cost-aware resource provisioning, and ML training trend predicting, to reduce the monetary cost and runtime of HPT processes. Our evaluations show that SpotTune can reduce the cost by up to 90% and achieve a 16.61x performance-cost rate improvement.

cs.DC↗

arXiv2020

S3ML: A Secure Serving System for Machine Learning Inference

We present S3ML, a secure serving system for machine learning inference in this paper. S3ML runs machine learning models in Intel SGX enclaves to protect users' privacy. S3ML designs a secure key management service to construct flexible privacy-preserving server clusters and proposes novel SGX-aware load balancing and scaling methods to satisfy users' Service-Level Objectives. We have implemented S3ML based on Kubernetes as a low-overhead, high-available, and scalable system. We demonstrate the system performance and effectiveness of S3ML through extensive experiments on a series of widely-used models.

cs.LG↗