arXiv subjects

Mingzhai Sun

Publications and source records attributed to Mingzhai Sun.

10 recordsLinked to original sources

arXiv2026

GRAFT: Grounded and Efficient Online Reinforcement Adaptation for Fine-Grained Robot Manipulation

Pretrained vision-language-action (VLA) policies provide strong priors for robot manipulation, yet adapting them online to fine-grained biomedical tasks remains challenging. Task success often hinges on subtle, view-dependent visual cues, while task-level rewards provide little guidance about which regions matter, making it difficult to learn task-relevant visual grounding from limited real-robot interaction. Online adaptation is further constrained by the computational cost of VLA inference and replay-based updates. We introduce GRAFT (Grounded Reinforcement Adaptation for Fast Task Learning), a framework for efficient online VLA adaptation through grounded perception. GRAFT uses region-level supervision to learn view-specific visual anchors that focus perception on task-relevant local cues without requiring region proposals at deployment. It further combines single-step action generation with cached visual-language prefix reuse to accelerate online learning. Across four biomedical manipulation tasks, GRAFT improves success rates by 32.5 percentage points under matched adaptation budgets, while reducing the computational overhead of online policy updates.

cs.RO↗

arXiv2026

THOR: A Theta-Gamma Hierarchical Oscillatory Reasoning Framework for Multi-hop QA

Multi-hop question answering requires retrieving and integrating evidence from multiple contexts. Despite the rapid progress of current research, multi-hop reasoning remains constrained by two persistent limitations: attention decay, where the model's focus on main question degrades as the reasoning chain grows, and error accumulation, where mistakes propagate across hops and compounds into final failure. Inspired by Theta-Gamma hierarchical oscillation which decouples global planning from local retrieval, enabling efficient attention transfer between hops and a verification and repair mechanism that interrupts the accumulation of errors in the wrong paths, we present THOR, a brain-inspired Theta-Gamma hierarchical oscillatory reasoning framework. Extensive comparative experiments and specific validation experiments on multi-hop QA benchmarks demonstrate that THOR improves answer accuracy and robustness while mitigating limitations, showcasing its generalization across different backbones.

cs.CL↗

arXiv2026

Keyframe-Guided Structured Rewards for Reinforcement Learning in Long-Horizon Laboratory Robotics

Long-horizon precision manipulation in laboratory automation, such as pipette tip attachment and liquid transfer, requires policies that respect strict procedural logic while operating in continuous, high-dimensional state spaces. However, existing approaches struggle with reward sparsity, multi-stage structural constraints, and noisy or imperfect demonstrations, leading to inefficient exploration and unstable convergence. We propose a Keyframe-Guided Reward Generation Framework that automatically extracts kinematics-aware keyframes from demonstrations, generates stage-wise targets via a diffusion-based predictor in latent space, and constructs a geometric progress-based reward to guide online reinforcement learning. The framework integrates multi-view visual encoding, latent similarity-based progress tracking, and human-in-the-loop reinforcement fine-tuning on a Vision-Language-Action backbone to align policy optimization with the intrinsic stepwise logic of biological protocols. Across four real-world laboratory tasks, including high-precision pipette attachment and dynamic liquid transfer, our method achieves an average success rate of 82% after 40--60 minutes of online fine-tuning. Compared with HG-DAgger (42%) and Hil-ConRFT (47%), our approach demonstrates the effectiveness of structured keyframe-guided rewards in overcoming exploration bottlenecks and providing a scalable solution for high-precision, long-horizon robotic laboratory automation.

cs.RO↗

arXiv2025

Intermediate Domain-guided Adaptation for Unsupervised Chorioallantoic Membrane Vessel Segmentation

The chorioallantoic membrane (CAM) model is a widely used in vivo platform for studying angiogenesis, especially in relation to tumor growth, drug delivery, and vascular biology.Since the topology and morphology of developing blood vessels is a key evaluation metric, accurate vessel segmentation is essential for quantitative analysis of angiogenesis. However, manual segmentation is extremely time-consuming, labor-intensive, and prone to inconsistency due to its subjective nature. Moreover, research on CAM vessel segmentation algorithms remains limited, and the lack of public datasets contributes to poor prediction performance. To address these challenges, we propose an innovative Intermediate Domain-guided Adaptation (IDA) method, which utilizes the similarity between CAM images and retinal images, along with existing public retinal datasets, to perform unsupervised training on CAM images. Specifically, we introduce a Multi-Resolution Asymmetric Translation (MRAT) strategy to generate intermediate images to promote image-level interaction. Then, an Intermediate Domain-guided Contrastive Learning (IDCL) module is developed to disentangle cross-domain feature representations. This method overcomes the limitations of existing unsupervised domain adaptation (UDA) approaches, which primarily concentrate on directly source-target alignment while neglecting intermediate domain information. Notably, we create the first CAM dataset to validate the proposed algorithm. Extensive experiments on this dataset show that our method outperforms compared approaches. Moreover, it achieves superior performance in UDA tasks across retinal datasets, highlighting its strong generalization capability. The CAM dataset and source codes are available at https://github.com/Light-47/IDA.

eess.IV↗

arXiv2025

BioMARS: A Multi-Agent Robotic System for Autonomous Biological Experiments

Large language models (LLMs) and vision-language models (VLMs) have the potential to transform biological research by enabling autonomous experimentation. Yet, their application remains constrained by rigid protocol design, limited adaptability to dynamic lab conditions, inadequate error handling, and high operational complexity. Here we introduce BioMARS (Biological Multi-Agent Robotic System), an intelligent platform that integrates LLMs, VLMs, and modular robotics to autonomously design, plan, and execute biological experiments. BioMARS uses a hierarchical architecture: the Biologist Agent synthesizes protocols via retrieval-augmented generation; the Technician Agent translates them into executable robotic pseudo-code; and the Inspector Agent ensures procedural integrity through multimodal perception and anomaly detection. The system autonomously conducts cell passaging and culture tasks, matching or exceeding manual performance in viability, consistency, and morphological integrity. It also supports context-aware optimization, outperforming conventional strategies in differentiating retinal pigment epithelial cells. A web interface enables real-time human-AI collaboration, while a modular backend allows scalable integration with laboratory hardware. These results highlight the feasibility of generalizable, AI-driven laboratory automation and the transformative role of language-based reasoning in biological research.

cs.RO↗

arXiv2025

Label-free Prediction of Vascular Connectivity in Perfused Microvascular Networks in vitro

Continuous monitoring and in-situ assessment of microvascular connectivity have significant implications for culturing vascularized organoids and optimizing the therapeutic strategies. However, commonly used methods for vascular connectivity assessment heavily rely on fluorescent labels that may either raise biocompatibility concerns or interrupt the normal cell growth process. To address this issue, a Vessel Connectivity Network (VC-Net) was developed for label-free assessment of vascular connectivity. To validate the VC-Net, microvascular networks (MVNs) were cultured in vitro and their microscopic images were acquired at different culturing conditions as a training dataset. The VC-Net employs a Vessel Queue Contrastive Learning (VQCL) method and a class imbalance algorithm to address the issues of limited sample size, indistinctive class features and imbalanced class distribution in the dataset. The VC-Net successfully evaluated the vascular connectivity with no significant deviation from that by fluorescence imaging. In addition, the proposed VC-Net successfully differentiated the connectivity characteristics between normal and tumor-related MVNs. In comparison with those cultured in the regular microenvironment, the averaged connectivity of MVNs cultured in the tumor-related microenvironment decreased by 30.8%, whereas the non-connected area increased by 37.3%. This study provides a new avenue for label-free and continuous assessment of organoid or tumor vascularization in vitro.

eess.IV↗

arXiv2014

Phase transitions during fruiting body formation in Myxococcus xanthus

The formation of a collectively moving group benefits individuals within a population in a variety of ways such as ultra-sensitivity to perturbation, collective modes of feeding, and protection from environmental stress. While some collective groups use a single organizing principle, others can dynamically shift the behavior of the group by modifying the interaction rules at the individual level. The surface-dwelling bacterium Myxococcus xanthus forms dynamic collective groups both to feed on prey and to aggregate during times of starvation. The latter behavior, termed fruiting-body formation, involves a complex, coordinated series of density changes that ultimately lead to three-dimensional aggregates comprising hundreds of thousands of cells and spores. This multi-step developmental process most likely involves several different single-celled behaviors as the population condenses from a loose, two-dimensional sheet to a three-dimensional mound. Here, we use high-resolution microscopy and computer vision software to spatiotemporally track the motion of thousands of individuals during the initial stages of fruiting body formation. We find that a combination of cell-contact-mediated alignment and internal timing mechanisms drive a phase transition from exploratory flocking, in which cell groups move rapidly and coherently over long distances, to a reversal-mediated localization into streams, which act as slow-spreading, quasi-one-dimensional nematic fluids. These observations lead us to an active liquid crystal description of the myxobacterial development cycle.

physics.bio-ph↗

arXiv2014

Myxococcus xanthus gliding motors are elastically coupled to the substrate as predicted by the focal adhesion model of gliding motility

Myxococcus xanthus is a model organism for studying bacterial social behaviors due to its ability to form complex multi-cellular structures. Knowledge of M. xanthus surface gliding motility and the mechanisms that coordinate it are critically important to our understanding of collective cell behaviors. Although the mechanism of gliding motility is still under investigation, recent experiments suggest that there are two possible mechanisms underlying force production for cell motility: the focal adhesion mechanism and the helical rotor mechanism which differ in the biophysics of the cell-substrate interactions. Whereas the focal adhesion model predicts an elastic coupling, the helical rotor model predicts a viscous coupling. Using a combination of computational modeling, imaging, and force microscopy, we find evidence for elastic coupling in support of the focal adhesion model. Using a biophysical model of the M. xanthus cell, we investigated how the mechanical interactions between cells are affected by interactions with the substrate. Comparison of modeling results with experimental data for cell-cell collision events pointed to a strong, elastic attachment between the cell and substrate. These results are robust to variations in the mechanical and geometrical parameters of the model. We then directly measured the motor-substrate coupling by monitoring the motion of optically trapped beads and find that motor velocity decreases exponentially with opposing load. At high loads, motor velocity approaches zero velocity asymptotically and motors remain bound to beads indicating a strong, elastic attachment.

physics.bio-ph↗

arXiv2012

Efficient Multiple Object Tracking Using Mutually Repulsive Active Membranes

Studies of social and group behavior in interacting organisms require high-throughput analysis of the motion of a large number of individual subjects. Computer vision techniques offer solutions to specific tracking problems, and allow automated and efficient tracking with minimal human intervention. In this work, we adopt the open active contour model to track the trajectories of moving objects at high density. We add repulsive interactions between open contours to the original model, treat the trajectories as an extrusion in the temporal dimension, and show applications to two tracking problems. The walking behavior of Drosophila is studied at different population density and gender composition. We demonstrate that individual male flies have distinct walking signatures, and that the social interaction between flies in a mixed gender arena is gender specific. We also apply our model to studies of trajectories of gliding Myxococcus xanthus bacteria at high density. We examine the individual gliding behavioral statistics in terms of the gliding speed distribution. Using these two examples at very distinctive spatial scales, we illustrate the use of our algorithm on tracking both short rigid bodies (Drosophila) and long flexible objects (Myxococcus xanthus). Our repulsive active membrane model reaches error rates better than $5\times 10^{-6}$ per fly per second for Drosophila tracking and comparable results for Myxococcus xanthus.

q-bio.QM↗

arXiv2011

Direct measurement of cell wall stress-stiffening and turgor pressure in live bacterial cells

We study intact and bulging Escherichia coli cells using atomic force microscopy to separate the contributions of the cell wall and turgor pressure to the overall cell stiffness. We find strong evidence of power-law stress-stiffening in the E. coli cell wall, with an exponent of 1.22 \pm 0.12, such that the wall is significantly stiffer in intact cells (E = 23 \pm 8 MPa and 49 \pm 20 MPa in the axial and circumferential directions) than in unpressurized sacculi. These measurements also indicate that the turgor pressure in living cells E. coli is 29 \pm 3 kPa.

physics.bio-ph↗