arXiv subjects

Zhiyuan Zhang

Publications and source records attributed to Zhiyuan Zhang.

At least 55 records · Page 3Linked to original sources

arXiv2026

Towards Realistic 3D Emission Materials: Dataset, Baseline, and Evaluation for Emission Texture Generation

3D texture generation is receiving increasing attention, as it enables the creation of realistic and aesthetic texture materials for untextured 3D meshes. However, existing 3D texture generation methods are limited to producing only a few types of non-emissive PBR materials (e.g., albedo, metallic maps and roughness maps), making them difficult to replicate highly popular styles, such as cyberpunk, failing to achieve effects like realistic LED emissions. To address this limitation, we propose a novel task, emission texture generation, which enables the synthesized 3D objects to faithfully reproduce the emission materials from input reference images. Our key contributions include: first, We construct the Objaverse-Emission dataset, the first dataset that contains 40k 3D assets with high-quality emission materials. Second, we propose EmissionGen, a novel baseline for the emission texture generation task. Third, we define detailed evaluation metrics for the emission texture generation task. Our results demonstrate significant potential for future industrial applications. Dataset will be available at https://github.com/yx345kw/EmissionGen.

cs.CV↗

arXiv2026

Bench2Drive-VL: Benchmarks for Closed-Loop Autonomous Driving with Vision-Language Models

With the rise of vision-language models (VLM), their application for autonomous driving (VLM4AD) has gained significant attention. Meanwhile, in autonomous driving, closed-loop evaluation has become widely recognized as a more reliable validation method than open-loop evaluation, as it can evaluate the performance of the model under cumulative errors and out-of-distribution inputs. However, existing VLM4AD benchmarks evaluate the model`s scene understanding ability under open-loop, i.e., via static question-answer (QA) dataset. This kind of evaluation fails to assess the VLMs performance under out-of-distribution states rarely appeared in the human collected datasets.To this end, we present Bench2Drive-VL, an extension of Bench2Drive that brings closed-loop evaluation to VLM-based driving, which introduces: (1) DriveCommenter, a closed-loop generator that automatically generates diverse, behavior-grounded question-answer pairs for all driving situations in CARLA,including severe off-route and off-road deviations previously unassessable in simulation. (2) A unified protocol and interface that allows modern VLMs to be directly plugged into the Bench2Drive closed-loop environment to compare with traditional agents. (3) A flexible reasoning and control framework, supporting multi-format visual inputs and configurable graph-based chain-of-thought execution. (4) A complete development ecosystem. Together, these components form a comprehensive closed-loop benchmark for VLM4AD. All codes and annotated datasets are open sourced.

cs.RO↗

arXiv2026

3D Dynamics-Aware Manipulation: Endowing Manipulation Policies with 3D Foresight

The incorporation of world modeling into manipulation policy learning has pushed the boundary of manipulation performance. However, existing efforts simply model the 2D visual dynamics, which is insufficient for robust manipulation when target tasks involve prominent depth-wise movement. To address this, we present a 3D dynamics-aware manipulation framework that seamlessly integrates 3D world modeling and policy learning. Three self-supervised learning tasks (current depth estimation, future RGB-D prediction, 3D flow prediction) are introduced within our framework, which complement each other and endow the policy model with 3D foresight. Extensive experiments on simulation and the real world show that 3D foresight can greatly boost the performance of manipulation policies without sacrificing inference speed. Code is available at https://github.com/Stardust-hyx/3D-Foresight.

cs.CV↗

arXiv2026

Composer 2 Technical Report

Composer 2 is a specialized model designed for agentic software engineering. The model demonstrates strong long-term planning and coding intelligence while maintaining the ability to efficiently solve problems for interactive use. The model is trained in two phases: first, continued pretraining to improve the model's knowledge and latent coding ability, followed by large-scale reinforcement learning to improve end-to-end coding performance through stronger reasoning, accurate multi-step execution, and coherence on long-horizon realistic coding problems. We develop infrastructure to support training in the same Cursor harness that is used by the deployed model, with equivalent tools and structure, and use environments that match real problems closely. To measure the ability of the model on increasingly difficult tasks, we introduce a benchmark derived from real software engineering problems in large codebases including our own. Composer 2 is a frontier-level coding model and demonstrates a process for training strong domain-specialized models. On our CursorBench evaluations the model achieves a major improvement in accuracy compared to previous Composer models (61.3). On public benchmarks the model scores 61.7 on Terminal-Bench and 73.7 on SWE-bench Multilingual in our harness, comparable to state-of-the-art systems.

cs.SE↗

arXiv2026

CityLens: Evaluating Large Vision-Language Models for Urban Socioeconomic Sensing

Understanding urban socioeconomic conditions through visual data is a challenging yet essential task for sustainable urban development and policy planning. In this work, we introduce \textit{CityLens}, a comprehensive benchmark designed to evaluate the capabilities of Large Vision-Language Models (LVLMs) in predicting socioeconomic indicators from satellite and street view imagery. We construct a multi-modal dataset covering a total of 17 globally distributed cities, spanning 6 key domains: economy, education, crime, transport, health, and environment, reflecting the multifaceted nature of urban life. Based on this dataset, we define 11 prediction tasks and utilize 3 evaluation paradigms: Direct Metric Prediction, Normalized Metric Estimation, and Feature-Based Regression. We benchmark 17 state-of-the-art LVLMs across these tasks. These make CityLens the most extensive socioeconomic benchmark to date in terms of geographic coverage, indicator diversity, and model scale. Our results reveal that while LVLMs demonstrate promising perceptual and reasoning capabilities, they still exhibit limitations in predicting urban socioeconomic indicators. CityLens provides a unified framework for diagnosing these limitations and guiding future efforts in using LVLMs to understand and predict urban socioeconomic patterns. The code and data are available at https://github.com/tsinghua-fib-lab/CityLens.

cs.AI↗

arXiv2026

(Dis)Proving Spectre Security with Speculation-Passing Style

Constant-time (CT) verification tools are commonly used for detecting potential side-channel vulnerabilities in cryptographic libraries. Recently, a new class of tools, called speculative constant-time (SCT) tools, has also been used for detecting potential Spectre vulnerabilities. In many cases, these SCT tools have emerged as liftings of CT tools. However, these liftings are seldom defined precisely and are almost never analyzed formally. The goal of this paper is to address this gap, by developing formal foundations for these liftings, and to demonstrate that these foundations can yield practical benefits. Concretely, we introduce a program transformation, coined Speculation-Passing Style (SPS), for reducing SCT verification to CT verification. Essentially, the transformation instruments the program with a new input that corresponds to attacker-controlled predictions and modifies the program to follow them. This approach is sound and complete, in the sense that a program is SCT if and only if its SPS transform is CT. Thus, we can leverage existing CT verification tools to prove SCT; we illustrate this by combining SPS with three standard methodologies for CT verification, namely reducing it to non-interference, assertion safety and dynamic taint analysis. We realize these combinations with three existing tools, EasyCrypt, BINSEC, and ctgrind, and we evaluate them on Kocher's benchmarks for Spectre-v1. Our results focus on Spectre-v1 in the standard CT leakage model; however, we also discuss applications of our method to other variants of Spectre and other leakage models.

cs.PL↗

arXiv2026

Super LiDAR Intensity for Robotic Perception

Conventionally, human intuition defines vision as a modality of passive optical sensing, relying on ambient light to perceive the environment. However, active optical sensing, which involves emitting and receiving signals, offers unique advantages by capturing both radiometric and geometric properties of the environment, independent of external illumination conditions. This work focuses on advancing active optical sensing using Light Detection and Ranging (LiDAR), which captures intensity data, enabling the estimation of surface reflectance that remains invariant under varying illumination. Such properties are crucial for robotic perception tasks, including detection, recognition, segmentation, and Simultaneous Localization and Mapping (SLAM). A key challenge with low-cost LiDARs lies in the sparsity of scan data, which limits their broader application. To address this limitation, this work introduces an innovative framework for generating dense LiDAR intensity images from sparse data, leveraging the unique attributes of non-repeating scanning LiDAR (NRS-LiDAR). We tackle critical challenges, including intensity calibration and the transition from static to dynamic scene domains, facilitating the reconstruction of dense intensity images in real-world settings. The key contributions of this work include a comprehensive dataset for LiDAR intensity image densification, a densification network tailored for NRS-LiDAR, and diverse applications such as loop closure and traffic lane detection using the generated dense intensity images. Experimental results validate the efficacy of the proposed approach, which successfully integrates computer vision techniques with LiDAR data processing, enhancing the applicability of low-cost LiDAR systems and establishing a novel paradigm for robotic vision via active optical sensing--LiDAR as a Camera.

cs.RO↗

arXiv2026

Learned Adaptive Mesh Generation

Elliptic Partial Differential Equations (PDEs) play a central role in computing the equilibrium conditions of physical problems (heat, gravitation, electrostatics, etc.). Efficient solutions to elliptic PDEs are also relevant to computer graphics since they encode global smoothness with local control leading to stable, well-behaved solutions. The Poisson equation is a linear elliptic PDE that serves as a prototypical candidate to assess newly-proposed solvers. Solving the Poisson equation on an arbitrary 3D domain, say a 3D scan of a turbine's blade, is computationally expensive and scales quadratically with discretization. Traditional workflows in research and industry exploit variants of the finite element method (FEM), but some key benefits of using Monte Carlo (MC) methods have been identified. Our key idea is to exploit a sparse and approximate solution (via FEM or MC) to the Poisson equation towards inferring an adaptive discretization in one shot. We achieve this by training a lightweight neural network that generalizes across shapes and boundary conditions. Our algorithm, Learned Adaptive Mesh Generation (LAMG), maps from a coarse solution to a sizing field that defines a local (adaptive) spatial resolution. This output space, rather than directly predicting a high-resolution solution, is a unique aspect of our approach. We use standard methods to generate tetrahedral meshes that respect the sizing field, and obtain the solution via one FEM computation on the adaptive mesh. That is, our neural network serves as a surrogate model of a computationally expensive method that requires multiple (iterative) FEM solves. We demonstrate the versatility, controllability, robustness and efficiency of LAMG via systematic experimentation.

cs.GR↗

arXiv2026

EquiForm: Noise-Robust SE(3)-Equivariant Policy Learning from 3D Point Clouds

Visual imitation learning with 3D point clouds has advanced robotic manipulation by providing geometry-aware, appearance-invariant observations. However, point cloud-based policies remain highly sensitive to sensor noise, pose perturbations, and occlusion-induced artifacts, which distort geometric structure and break the equivariance assumptions required for robust generalization. Existing equivariant approaches primarily encode symmetry constraints into neural architectures, but do not explicitly correct noise-induced geometric deviations or enforce equivariant consistency in learned representations. We introduce EquiForm, a noise-robust SE(3)-equivariant policy learning framework for point cloud-based manipulation. EquiForm formalizes how noise-induced geometric distortions lead to equivariance deviations in observation-to-action mappings, and introduces a geometric denoising module to restore consistent 3D structure under noisy or incomplete observations. In addition, we propose a contrastive equivariant alignment objective that enforces representation consistency under both rigid transformations and noise perturbations. Built upon these components, EquiForm forms a flexible policy learning pipeline that integrates noise-robust geometric reasoning with modern generative models. We evaluate EquiForm on 16 simulated tasks and 4 real-world manipulation tasks across diverse objects and scene layouts. Compared to state-of-the-art point cloud imitation learning methods, EquiForm achieves an average improvement of 17.2% in simulation and 28.1% in real-world experiments, demonstrating strong noise robustness and spatial generalization.

cs.RO↗

arXiv2026

ManiFeel: Benchmarking and Understanding Visuotactile Manipulation Policy Learning

Supervised visuomotor policies have shown strong performance in robotic manipulation but often struggle in tasks with limited visual inputs, such as operations in confined spaces and dimly lit environments, or tasks requiring precise perception of object properties and environmental interactions. In such cases, tactile feedback becomes essential for manipulation. While the rapid progress of supervised visuomotor policies has benefited greatly from high-quality, reproducible simulation benchmarks in visual imitation, the visuotactile domain still lacks a similarly comprehensive and reliable benchmark for large-scale and rigorous evaluation. To address this, we introduce ManiFeel, a reproducible and scalable simulation benchmark designed to systematically study supervised visuotactile policy learning. ManiFeel offers a diverse suite of contact-rich and visually challenging manipulation tasks, a modular evaluation pipeline spanning sensing modalities, tactile representations, and policy architectures, as well as real-world validation. Through extensive experiments, ManiFeel demonstrates how tactile sensing enhances policy performance across diverse manipulation scenarios, ranging from precise contact-driven operations to visually constrained settings. In addition, the results reveal task-dependent strengths of different tactile modalities and identify key design principles and open challenges for robust visuotactile policy learning. Real-world evaluations further confirm that ManiFeel provides a reliable and meaningful foundation for benchmarking and future visuotactile policy development. To foster reproducibility and future research, we will release our codebase, datasets, training logs, and pretrained checkpoints, aiming to accelerate progress toward generalizable visuotactile policy learning and manipulation.

cs.RO↗

arXiv2025

Spatial Retrieval Augmented Autonomous Driving

Existing autonomous driving systems rely on onboard sensors (cameras, LiDAR, IMU, etc) for environmental perception. However, this paradigm is limited by the drive-time perception horizon and often fails under limited view scope, occlusion or extreme conditions such as darkness and rain. In contrast, human drivers are able to recall road structure even under poor visibility. To endow models with this ``recall" ability, we propose the spatial retrieval paradigm, introducing offline retrieved geographic images as an additional input. These images are easy to obtain from offline caches (e.g, Google Maps or stored autonomous driving datasets) without requiring additional sensors, making it a plug-and-play extension for existing AD tasks. For experiments, we first extend the nuScenes dataset with geographic images retrieved via Google Maps APIs and align the new data with ego-vehicle trajectories. We establish baselines across five core autonomous driving tasks: object detection, online mapping, occupancy prediction, end-to-end planning, and generative world modeling. Extensive experiments show that the extended modality could enhance the performance of certain tasks. We will open-source dataset curation code, data, and benchmarks for further study of this new autonomous driving paradigm.

cs.CV↗

arXiv2025

MoReFun: Past-Movement Guided Motion Representation Learning for Future Motion Prediction and Understanding

3D human motion prediction aims to generate coherent future motions from observed sequences, yet existing end-to-end regression frameworks often fail to capture complex dynamics and tend to produce temporally inconsistent or static predictions-a limitation rooted in representation shortcutting, where models rely on superficial cues rather than learning meaningful motion structure. We propose a two-stage self-supervised framework that decouples representation learning from prediction. In the pretraining stage, the model performs unified past-future self-reconstruction, reconstructing the past sequence while recovering masked joints in the future sequence under full historical guidance. A velocity-based masking strategy selects highly dynamic joints, forcing the model to focus on informative motion components and internalize the statistical dependencies between past and future states without regression interference. In the fine-tuning stage, the pretrained model predicts the entire future sequence, now treated as fully masked, and is further equipped with a lightweight future-text prediction head for joint optimization of low-level motion prediction and high-level motion understanding. Experiments on Human3.6M, 3DPW, and AMASS show that our method reduces average prediction errors by 8.8% over state-of-the-art methods while achieving competitive future-motion understanding performance compared to LLM-based models. Code is available at: https://github.com/JunyuShi02/MoReFun

cs.CV↗

arXiv2025

EmbryoDiff: A Conditional Diffusion Framework with Multi-Focal Feature Fusion for Fine-Grained Embryo Developmental Stage Recognition

Identification of fine-grained embryo developmental stages during In Vitro Fertilization (IVF) is crucial for assessing embryo viability. Although recent deep learning methods have achieved promising accuracy, existing discriminative models fail to utilize the distributional prior of embryonic development to improve accuracy. Moreover, their reliance on single-focal information leads to incomplete embryonic representations, making them susceptible to feature ambiguity under cell occlusions. To address these limitations, we propose EmbryoDiff, a two-stage diffusion-based framework that formulates the task as a conditional sequence denoising process. Specifically, we first train and freeze a frame-level encoder to extract robust multi-focal features. In the second stage, we introduce a Multi-Focal Feature Fusion Strategy that aggregates information across focal planes to construct a 3D-aware morphological representation, effectively alleviating ambiguities arising from cell occlusions. Building on this fused representation, we derive complementary semantic and boundary cues and design a Hybrid Semantic-Boundary Condition Block to inject them into the diffusion-based denoising process, enabling accurate embryonic stage classification. Extensive experiments on two benchmark datasets show that our method achieves state-of-the-art results. Notably, with only a single denoising step, our model obtains the best average test performance, reaching 82.8% and 81.3% accuracy on the two datasets, respectively.

cs.CV↗

arXiv2025

Privacy-Preserving Retrieval-Augmented Generation with Differential Privacy

With the recent remarkable advancement of large language models (LLMs), there has been a growing interest in utilizing them in the domains with highly sensitive data that lies outside their training data. For this purpose, retrieval-augmented generation (RAG) is particularly effective -- it assists LLMs by directly providing relevant information from the external knowledge sources. However, without extra privacy safeguards, RAG outputs risk leaking sensitive information from the external data source. In this work, we explore RAG under differential privacy (DP), a formal guarantee of data privacy. The main challenge with differentially private RAG is how to generate long accurate answers within a moderate privacy budget. We address this by proposing an algorithm that smartly spends privacy budget only for the tokens that require the sensitive information and uses the non-private LLM for other tokens. Our extensive empirical evaluations reveal that our algorithm outperforms the non-RAG baseline under a reasonable privacy budget of $ε\approx 10$ across different models and datasets.

cs.CR↗

arXiv2025

Private-RAG: Answering Multiple Queries with LLMs while Keeping Your Data Private

Retrieval-augmented generation (RAG) enhances large language models (LLMs) by retrieving documents from an external corpus at inference time. When this corpus contains sensitive information, however, unprotected RAG systems are at risk of leaking private information. Prior work has introduced differential privacy (DP) guarantees for RAG, but only in single-query settings, which fall short of realistic usage. In this paper, we study the more practical multi-query setting and propose two DP-RAG algorithms. The first, MURAG, leverages an individual privacy filter so that the accumulated privacy loss only depends on how frequently each document is retrieved rather than the total number of queries. The second, MURAG-ADA, further improves utility by privately releasing query-specific thresholds, enabling more precise selection of relevant documents. Our experiments across multiple LLMs and datasets demonstrate that the proposed methods scale to hundreds of queries within a practical DP budget ($\varepsilon\approx10$), while preserving meaningful utility.

cs.LG↗

arXiv2025

Canonical Policy: Learning Canonical 3D Representation for SE(3)-Equivariant Policy

Visual Imitation learning has achieved remarkable progress in robotic manipulation, yet generalization to unseen objects, scene layouts, and camera viewpoints remains a key challenge. Recent advances address this by using 3D point clouds, which provide geometry-aware, appearance-invariant representations, and by incorporating equivariance into policy architectures to exploit spatial symmetries. However, existing equivariant approaches often lack interpretability and rigor due to unstructured integration of equivariant components. We introduce canonical policy, a principled framework for 3D equivariant imitation learning that unifies 3D point cloud observations under a canonical representation. We first establish a theory of 3D canonical representations, enabling equivariant observation-to-action mappings by grouping both seen and novel point clouds to a canonical representation. We then propose a flexible policy learning pipeline that leverages geometric symmetries from canonical representation and the expressiveness of modern generative models. We validate canonical policy on 12 diverse simulated tasks and 4 real-world manipulation tasks across 16 configurations, involving variations in object color, shape, camera viewpoint, and robot platform. Compared to state-of-the-art imitation learning policies, canonical policy achieves an average improvement of 18.0% in simulation and 39.7% in real-world experiments, demonstrating superior generalization capability and sample efficiency. For more details, please refer to the project website: https://zhangzhiyuanzhang.github.io/cp-website/.

cs.RO↗

arXiv2025

Broad nonlocal spectrum in the Pb-InSb hybrid three terminals for potential realization of Kitaev chains

Hybrid superconductor-semiconductor(SC-SM) nanowires remain one of the foremost platforms for engineering topological superconductivity and Majorana zero modes(MZMs) towards fault-tolerant topological qubits, especially with the rapid development of artificial Kitaev chains. In contrast to the widely used aluminum(Al)-based hybrids, lead(Pb) offers a bulk superconducting gap of ~1.4meV and a critical temperature of ~7.2K, giving rise to a proximity-induced gap that is roughly five times larger than that obtained with Al. Here we present the first three-terminal Pb-hybrid devices and perform nonlocal differential-conductance spectroscopy on this platform. The nonlocal measurement simultaneously resolves a dual-gap feature of the parent Pb gap and the large, hard, gate-tunable induced superconducting gap, distinguished by a switch between electron- and hole-like dissipation processes. Within the induced gap we observe several types of Andreev bound states(ABSs) that undergo singlet-doublet transitions. Moreover, by tuning gate voltages we achieve gate-controlled resonating sign reversals of the nonlocal conductance, identifying three distinct regimes that correspond to different configurations of quantum-dot(QD) resonances(single-resonance, double-resonance, and series-resonance). Finally, the coupling between ABSs and QDs also present and can be modulated from the weak- to strong-coupling limit, indicating the feasibility of realizing the artificial Kitaev chains. Crucially, the robust nonlocal signatures persist up to temperatures(~1K) far above the operating temperature of Al-based devices thanks to the unusually large induced gap, thereby widening the accessible parameter space greatly and underscoring the suitability of Pb-based hybrids for implementing warm temperature artificial Kitaev chains and the topological quantum devices protected by a substantially larger topological gap.

quant-ph↗

arXiv2025

Magnetic-Field Control of Tomonaga-Luttinger Liquids in Ta2Pd3Te5 Edge States

Ta2Pd3Te5 is a quasi-one-dimensional transition-metal telluride whose heavy atoms endow the material with strong spin-orbit coupling, while the Fermi level inside the bulk gap makes the low-energy electronic structure highly tunable.Theory and early experiments have already identified a wealth of emergent phases in this platform: an excitonic insulator driven by electron-hole binding, a second-order topological insulator protected by crystalline symmetry, a potential topological-protected quantum-spin-Hall edge, and proximity-induced edge supercurrents when coupled to a conventional s-wave superconductor. These properties make it a promising platform for hosting Majorana zero modes and quantum computation, provided that time-reversal symmetry can be broken by a Zeeman gap. In this work, we demonstrate that the one-dimensional edge channels of exfoliated Ta2Pd3Te5 host a robust and tunable Tomonaga-Luttinger liquid by electrostatic gating because it shifts the chemical potential across the bulk gap without changing the gap size. More importantly, the application of a magnetic field introduces a Zeeman gap that systematically increases the TLL power-law exponent alpha. Furthermore, rotating the field reveals a pronounced twofold anisotropy--alpha is maximal for a field parallel to the edge and minimal for a perpendicular orientation--originating from an orientation-dependent edge g-factor that is likely amplified by quantum-confinement-induced orbital-angular-moment quenching. The existence of gate-tunable edge supercurrents together with the field-controlled Zeeman gap provides a direct route to break time-reversal symmetry in a particle-hole-symmetric superconducting gap and thus to engineer a topological superconducting phase, paving the way towards Majorana-based quantum devices.

cond-mat.mes-hall↗