arXiv subjects

Zizhe Wang

Publications and source records attributed to Zizhe Wang.

16 recordsLinked to original sources

arXiv2026

AI for Science in Space: Orbital Laboratories, Planetary Fieldwork, and Embodied Autonomy

Artificial intelligence for science (AI4S) is extending from interpretation of existing data to the design and execution of experiments. Space offers distinctive scientific opportunities while restricting physical support. Microgravity and extraterrestrial environments expose phenomena that terrestrial experiments cannot fully reproduce; finite samples, constrained resources, and limited opportunities for intervention complicate sustained investigation. This critical review examines orbital laboratories and planetary in situ science across biology, materials, fluid and thermal physics, and adaptive exploration. We distinguish learning from space-derived data, selecting scientific actions, executing them, and preserving experimental validity. Terrestrial self-driving laboratories provide complementary evidence on optimization, dedicated automation, robotic manipulation, and the support needed for unattended operation. The synthesis connects scientific gains already demonstrated by bounded AI systems to gaps in observation, executable actions, and continuity of valid evidence. Space robots can expand the attainable scope of autonomous science when necessary interventions exceed validated instrument capabilities and timely human support is unavailable. Their value must be established against dedicated mechanisms, redundancy, and human-supported operation. We develop a framework connecting sample history, scientific decisions, physical execution, and claim-specific validity. A worked orbital design, a planetary extension, and matched evaluation schemes separate the contributions of sensing, decision policies, and physical actions. The resulting agenda measures autonomy through valid scientific return, task coverage, intervention costs, and evidence integrity across complete campaigns.

cs.RO↗

arXiv2026

AstroBench: Benchmarking Free-Floating Capture and Space Robotic Manipulation

Capturing a loose object inside a spacecraft requires a robot to approach a target that can translate and rotate, and whose motion changes when contact is made. These conditions motivate evaluation beyond manipulation of supported objects. We introduce AstroBench, an open benchmark for free-floating capture and space robotic manipulation with a fixed-base arm. The task suite combines four initial-motion tracks with sample-storage operations based on MELFI hardware and cap and captive-screw servicing based on the Robotic Refueling Mission. The two workcells provide nine task definitions and 29 task-target combinations. A metadata-filtered Objaverse inventory supplies 8,414 object instances for asset expansion, alongside five cabin visual contexts and versioned task instructions. We report two acquisition-focused experiments using existing checkpoints. A frozen $π_{0.5}$ DROID joint-position policy captures 2 of 200 static targets and 4 of 200 moving targets across twenty object identities. A cube-specific ACT policy captures 8 of 10 static targets and 6 of 10 moving targets in a separate experiment. Each experiment retains its recorded observation, action and capture configuration. The results document difficult direct transfer for the tested generalist checkpoint and successful task-specific capture behavior, while identifying concrete settings for studying motion-aware perception and contact-sensitive control. Code, task definitions and editable workcell assets are available at https://github.com/AstroUpLab/AstroBench.

cs.RO↗

arXiv2026

Compressing Onboard Language Models for Space Robots under a Behavioral-Fidelity Requirement

Language models are being explored as the interface through which space robots interpret operator commands and select among verified skills. Running them onboard calls for low-bit compression, and flight use adds requirements that generic post-training quantization leaves implicit: a fixed memory budget and storage format, a static artifact under configuration control, and behavior close to the full-precision model validated on the ground. We treat compression as a ground-segment step and study QSelect, which composes one static low-bit model from compatible AWQ, GPTQ, and AutoRound checkpoints by assigning each linear module to one quantizer family and scoring every substitution at the final next-token distribution. The default fidelity mode needs no labels or tuned thresholds and accepts a substitution only when the forward KL divergence from the BF16 reference decreases; an optional target-adapted mode adds answer likelihood on a small labeled set. On DeepSeek-R1-Distill-Qwen-7B, Llama-3.1-8B, and Llama-3.2-3B at 3 and 4 bits, composed models reach lower mean held-out perplexity than every single-family model in all six settings, and the fidelity mode lowers perplexity and KL further. In a matched scaffold, final-output scoring gives lower held-out KL than representation-similarity scoring in all 12 paired runs. With labels from training or development splits of the evaluated task families, the target-adapted mode raises 3-bit pooled accuracy on the two Llama models by 4.3 and 4.5 points over the best single-family model. At 3 bits, unquantized embedding and output layers make up 43-46% of the packed artifact. All evaluations use general-domain benchmarks on a ground GPU; space-domain tasks and flight hardware are left to future work.

cs.RO↗

arXiv2026

Foundation Models for Autonomous Space Robotics: Architectures, Physical Adaptation, and Validation

Space robots must interpret tasks and act despite unfamiliar dynamics, difficult sensing conditions, and limited opportunities for human intervention. Foundation models offer reusable perceptual and behavioral knowledge, but their value depends on how that knowledge reaches physical execution. This critical review follows space-specific requirements through perception, prediction, planning, and action to deployment and validation, emphasizing orbital manipulation and proximity operations. Terrestrial comparisons establish benefits of particular pretraining stages and model interfaces; their transfer to space dynamics and constrained onboard operation remains insufficiently tested. Event cameras offer timely motion observations across strong illumination contrasts, while spiking neural networks provide possible routes to sparse local computation. We examine how sensor degradation, state estimation and computing choices affect feedback and recovery. Higher action throughput need not yield fresher feedback; lower processor power can also coexist with higher energy per inference. We organize validation as an iterative sim-to-ground-to-space pathway, connecting physical tests with telemetry-based diagnosis and closed-loop revalidation. A satellite-capture example links observations to approach, contact and recovery, while a corresponding protocol separates generalization from adaptation with limited target-domain interaction. These comparisons identify what models already contribute and the physical evidence needed to support autonomous space operation.

cs.RO↗

arXiv2026

Scoring, Reasoning, and Selecting the Best! Ensembling Large Language Models via a Peer-Review Process

We propose LLM-PeerReview, an unsupervised LLM Ensemble method that selects the most ideal response from multiple LLM-generated candidates for each query, harnessing the collective wisdom of multiple models with diverse strengths. LLM-PeerReview is built on a novel, peer-review-inspired framework that offers a transparent and interpretable mechanism, while remaining fully unsupervised for flexible adaptability and generalization. Specifically, it operates in three stages: For scoring, we use the emerging LLM-as-a-Judge technique to evaluate each response by reusing multiple LLMs at hand; For reasoning, we can apply a straightforward averaging strategy or a principled graphical model-based truth inference algorithm to aggregate multiple scores to produce a final score for each response; Finally, the highest-scoring response is selected as the best ensemble output. LLM-PeerReview is conceptually simple and empirically powerful. Our results across four datasets show that the two variants of the proposed approach outperform the advanced model Smoothie-Global by 6.9% and 7.3% points, cross diverse task types including factual recall QA, math reasoning, and instruction following. Notably, we also establish a carefully curated benchmark suite for LLM Ensemble, integrating 12 methods across four classic datasets and three task families, all evaluated under a rigorous and consistent protocol. We hope this repository will help researchers reproduce the LLM Ensemble baselines.

cs.CL↗

arXiv2026

Beyond Executable Models: The Pufibara Agent Harness and the Modelica Agent Workflow Benchmark for Physical System Modeling

AI agents are increasingly used for simulation-driven engineering. Physical system modeling presents different requirements from general-purpose code generation in software engineering, because correctness depends not only on syntax and executability but also on physical consistency and scenario-dependent behavior. We study this challenge in Modelica, an equation-based modeling language in which a model may compile and simulate while still violating its intended physics or engineering requirements. Across successive revisions, an agent may lose track of requirements or rely on simulation evidence produced by an outdated candidate. To address this challenge, we present Pufibara, an agent harness that maintains persistent engineering state across revisions, associates execution and simulation evidence with the candidate that produced it, and makes submission an explicit agent action. To evaluate end-to-end Modelica agent workflows, we also propose a source-grounded method for constructing realistic and independently evaluable tasks. We use this method to build the 232-task Modelica Agent Workflow Benchmark, spanning Model Repair, Model Generation, and Model Tuning. Each submitted candidate is scored by a benchmark-owned evaluator outside the agent loop. We compare Pufibara with Claude Code as complete harnesses under two matched large language model (LLM) backends. With DeepSeek v4 Flash, Pufibara passes 202 tasks, compared with 185 for Claude Code. With Claude Sonnet 5, Pufibara passes 202 tasks, compared with 187 for Claude Code. Under the repository-reported token accounting, Pufibara records 76.4%-82.5% lower logical-token totals. Its sequential runtime is 6.1%-58.4% lower. These findings show that, even under matched LLM backends, complete agent harnesses can differ substantially in both task success and resource use for physical system modeling.

cs.SE↗

arXiv2026

End-to-end Conditional Diffusion for Realistic and Controllable Visual Traffic Scenario Generation

Generating closed-loop traffic scenarios that are both realistic and controllable is crucial for evaluating autonomous driving systems, especially under rare safety-critical interactions. Existing learning-based methods often struggle to balance controllability and realism, offering either limited fine-grained control over traffic behavior or controllable scenarios at the expense of behavioral plausibility. This paper presents E2E-CDiff, an end-to-end conditional diffusion framework for controllable and realistic scenario generation. Conditioned on front-view visual observations, E2E-CDiff jointly denoises future motion states and executable low-level controls for route-interacting background vehicles. This unified state-action generation mitigates the planning-control mismatch in conventional two-stage trajectory-then-controller pipelines. Differentiable guidance further regulates speed, enforces drivable-area compliance, and supports collision-avoidance or collision-seeking behaviors, enabling both naturalistic and safety-critical scenario generation. Experiments on Bench2Drive show that E2E-CDiff achieves a favorable controllability-realism trade-off compared with representative reinforcement- and imitation-learning baselines, while its collision-guided variant induces challenging interactions across multiple autonomous driving systems. E2E-CDiff also performs competitively as a learning-based ego planner, demonstrating the generality of end-to-end state-action diffusion.

cs.RO↗

arXiv2026

Beyond Scaling: Assessing Strategic Reasoning and Rapid Decision-Making Capability of LLMs in Zero-sum Environments

Large Language Models (LLMs) have achieved strong performance on static reasoning benchmarks, yet their effectiveness as interactive agents operating in adversarial, time-sensitive environments remains poorly understood. Existing evaluations largely treat reasoning as a single-shot capability, overlooking the challenges of opponent-aware decision-making, temporal constraints, and execution under pressure. This paper introduces Strategic Tactical Agent Reasoning (STAR) Benchmark, a multi-agent evaluation framework that assesses LLMs through 1v1 zero-sum competitive interactions, framing reasoning as an iterative, adaptive decision-making process. STAR supports both turn-based and real-time settings, enabling controlled analysis of long-horizon strategic planning and fast-paced tactical execution within a unified environment. Built on a modular architecture with a standardized API and fully implemented execution engine, STAR facilitates reproducible evaluation and flexible task customization. To move beyond binary win-loss outcomes, we introduce a Strategic Evaluation Suite that assesses not only competitive success but also the quality of strategic behavior, such as execution efficiency and outcome stability. Extensive pairwise evaluations reveal a pronounced strategy-execution gap: while reasoning-intensive models dominate turn-based settings, their inference latency often leads to inferior performance in real-time scenarios, where faster instruction-tuned models prevail. These results show that strategic intelligence in interactive environments depends not only on reasoning depth, but also on the ability to translate plans into timely actions, positioning STAR as a principled benchmark for studying this trade-off in competitive, dynamic settings.

cs.CV↗

arXiv2025

Noise Diffusion for Enhancing Semantic Faithfulness in Text-to-Image Synthesis

Diffusion models have achieved impressive success in generating photorealistic images, but challenges remain in ensuring precise semantic alignment with input prompts. Optimizing the initial noisy latent offers a more efficient alternative to modifying model architectures or prompt engineering for improving semantic alignment. A latest approach, InitNo, refines the initial noisy latent by leveraging attention maps; however, these maps capture only limited information, and the effectiveness of InitNo is highly dependent on the initial starting point, as it tends to converge on a local optimum near this point. To this end, this paper proposes leveraging the language comprehension capabilities of large vision-language models (LVLMs) to guide the optimization of the initial noisy latent, and introduces the Noise Diffusion process, which updates the noisy latent to generate semantically faithful images while preserving distribution consistency. Furthermore, we provide a theoretical analysis of the condition under which the update improves semantic faithfulness. Experimental results demonstrate the effectiveness and adaptability of our framework, consistently enhancing semantic alignment across various diffusion models. The code is available at https://github.com/Bomingmiao/NoiseDiffusion.

cs.CV↗

arXiv2025

Towards Understanding How Knowledge Evolves in Large Vision-Language Models

Large Vision-Language Models (LVLMs) are gradually becoming the foundation for many artificial intelligence applications. However, understanding their internal working mechanisms has continued to puzzle researchers, which in turn limits the further enhancement of their capabilities. In this paper, we seek to investigate how multimodal knowledge evolves and eventually induces natural languages in LVLMs. We design a series of novel strategies for analyzing internal knowledge within LVLMs, and delve into the evolution of multimodal knowledge from three levels, including single token probabilities, token probability distributions, and feature encodings. In this process, we identify two key nodes in knowledge evolution: the critical layers and the mutation layers, dividing the evolution process into three stages: rapid evolution, stabilization, and mutation. Our research is the first to reveal the trajectory of knowledge evolution in LVLMs, providing a fresh perspective for understanding their underlying mechanisms. Our codes are available at https://github.com/XIAO4579/Vlm-interpretability.

cs.CV↗

arXiv2025

AdvLogo: Adversarial Patch Attack against Object Detectors based on Diffusion Models

With the rapid development of deep learning, object detectors have demonstrated impressive performance; however, vulnerabilities still exist in certain scenarios. Current research exploring the vulnerabilities using adversarial patches often struggles to balance the trade-off between attack effectiveness and visual quality. To address this problem, we propose a novel framework of patch attack from semantic perspective, which we refer to as AdvLogo. Based on the hypothesis that every semantic space contains an adversarial subspace where images can cause detectors to fail in recognizing objects, we leverage the semantic understanding of the diffusion denoising process and drive the process to adversarial subareas by perturbing the latent and unconditional embeddings at the last timestep. To mitigate the distribution shift that exposes a negative impact on image quality, we apply perturbation to the latent in frequency domain with the Fourier Transform. Experimental results demonstrate that AdvLogo achieves strong attack performance while maintaining high visual quality.

cs.CV↗

arXiv2024

An Efficient Framework for Enhancing Discriminative Models via Diffusion Techniques

Image classification serves as the cornerstone of computer vision, traditionally achieved through discriminative models based on deep neural networks. Recent advancements have introduced classification methods derived from generative models, which offer the advantage of zero-shot classification. However, these methods suffer from two main drawbacks: high computational overhead and inferior performance compared to discriminative models. Inspired by the coordinated cognitive processes of rapid-slow pathway interactions in the human brain during visual signal recognition, we propose the Diffusion-Based Discriminative Model Enhancement Framework (DBMEF). This framework seamlessly integrates discriminative and generative models in a training-free manner, leveraging discriminative models for initial predictions and endowing deep neural networks with rethinking capabilities via diffusion models. Consequently, DBMEF can effectively enhance the classification accuracy and generalization capability of discriminative models in a plug-and-play manner. We have conducted extensive experiments across 17 prevalent deep model architectures with different training methods, including both CNN-based models such as ResNet and Transformer-based models like ViT, to demonstrate the effectiveness of the proposed DBMEF. Specifically, the framework yields a 1.51\% performance improvement for ResNet-50 on the ImageNet dataset and 3.02\% on the ImageNet-A dataset. In conclusion, our research introduces a novel paradigm for image classification, demonstrating stable improvements across different datasets and neural networks. The code is available at https://github.com/ChunXiaostudy/DBMEF.

cs.CV↗

arXiv2021

Coupling innovation method and feasibility analysis of garbage classification

In order to solve the recent defect in garbage classification - including low level of intelligence, low accuracy and high cost of equipment, this paper presents a series of methods in identification and judgment in intelligent garbage classification, including a material identification based on thermal principle and non-destructive laser irradiation, another material identification based on optical diffraction and phase analysis, a profile identification which utilizes a scenery thermal image after PCA and histogram correction, another profile identification which utilizes computer vision with innovated data sets and algorithms. Combining AHP and Bayesian formula, the paper innovates a coupling algorithm which helps to make a comprehensive judgment of the garbage sort, based on the material and profile identification. This paper also proposes a method for real-time space measurement of garbage cans, which based on the characteristics of air as fluid, and analyses the functions of air cleaning and particle disposing. Instead of the single use of garbage image recognition, this paper provides a comprehensive method to judge the garbage sort by material and profile identifications, which greatly enhancing the accuracy and intelligence in garbage classification.

cs.LG↗

arXiv2020

Chiral state conversion in a levitated micromechanical oscillator with in situ control of parameter loops

Physical systems with gain and loss can be described by a non-Hermitian Hamiltonian, which is degenerated at the exceptional points (EPs). Many new and unexpected features have been explored in the non-Hermitian systems with a great deal of recent interest. One of the most fascinating features is that, chiral state conversion appears when one EP is encircled dynamically. Here, we propose an easy-controllable levitated microparticle system that carries a pair of EPs and realize slow evolution of the Hamiltonian along loops in the parameter plane. Utilizing the controllable rotation angle, gain and loss coefficients, we can control the structure, size and location of the loops in situ. We demonstrate that, under the joint action of topological structure of energy surfaces and nonadiabatic transitions (NATs), the chiral behavior emerges both along a loop encircling an EP and even along a straight path away from the EP. This work broadens the range of parameter space for the chiral state conversion, and proposes a useful platform to explore the interesting properties of exceptional points physics.

quant-ph↗

arXiv2019

Room temperature test of the Continuous Spontaneous Localization model using a levitated micro-oscillator

The Continuous Spontaneous Localization (CSL) model predicts a tiny break of energy conservation via a weak stochastic force acting on physical systems, which triggers the collapse of the wave function. Mechanical oscillators are a natural way to test such a force; in particular levitated micro-mechanical oscillator has been recently proposed to be an ideal system. We report a proof-of-principle experiment with a micro-oscillator generated by a micro-sphere diamagnetically levitated in a magneto-gravitational trap under high vacuum. Due to the ultra-low mechanical dissipation, the oscillator provides a new upper bound on the CSL collapse rate, which gives an improvement of two orders of magnitude over the previous bounds in the same frequency range, and partially reaches the enhanced collapse rate suggested by Adler. Although being performed at room temperature, our experiment has already exhibits advantages over those operating at low temperatures previously reported. Our results experimentally show the potential of magneto-gravitational levitated mechanical oscillator as a promising method for testing collapse model. Further improvements in cryogenic experiments are discussed.

quant-ph↗

arXiv2019

Manipulation of a Micro-object Using Topological Hydrodynamic Tweezers

Manipulating micro-scale object plays paramount roles in a wide range of fundamental researches and applications. At micro-scale, various methods have been developed in the past decades, including optical, electric, magnetic, aerodynamic and acoustic methods. However, these non-contact forces are susceptible to external disturbance, and so finding a way to make micro-scale object manipulation immune to external perturbations is challenging and remains elusive. Here we demonstrate a method based on new trapping mechanism to manipulate micro-scale object in a gas flow at ambient conditions. We first show that the micro-droplet is entrapped into a trapping ring constructed by a particular toroidal vortex. The vortex works as tweezers to control the position of the micro-droplet. We then show that the micro-droplet can be transported along the trapping ring. By virtue of the topological character of the gas flow, the transport path is able to bypass external strong perturbations automatically. We further demonstrate a topological transfer process of the micro-droplet between two hydrodynamic tweezers. Our method provides an integrated toolbox to manipulate a micro-scale object, with an intrinsic mechanism that protects the target object from external disturbances.

physics.flu-dyn↗