arXiv subjects

Yingjie Xu

Publications and source records attributed to Yingjie Xu.

At least 19 recordsLinked to original sources

arXiv2026

WorldLines: Benchmarking and Modeling Long-Horizon Stateful Embodied Agents

To assist humans over extended periods in real homes, embodied agents must remember user routines, world states, and past interactions. Existing long-term memory benchmarks mainly evaluate language-centric retrieval and question answering, while embodied benchmarks often focus on short-horizon task execution without testing long-term memory use in dynamic environments. We introduce WorldLines, a project-driven benchmark for long-horizon embodied household assistance. It constructs temporally extended household traces with dialogues, actions, execution feedback, object and device state changes, and converts them into evidence-linked samples for Memory QA and Embodied Task Planning. We further propose ObsMem, an observer-grounded memory framework that maintains visibility-aware memories and action-native state trails for state-aware decisions. Experiments reveal persistent challenges in partial observability, overwritten world states, and translating long-term memory into embodied plans, while ObsMem offers a stronger reference architecture for this setting.

cs.AI↗

arXiv2026

SeisEvo: Evolution of Seismic Data Reconstruction Algorithms by Agents

Classical seismic data reconstruction relies on manually designed structural priors and iterative operators, whose coupled design space is far larger than manual trial and error can explore systematically. Deep-learning methods encode the reconstruction rules in learned weights rather than in an explicit operator that can be inspected and modified. We propose SeisEvo (Seismic Algorithm Evolution), which does not optimize a single reconstruction result but searches for the algorithm that produces it. Starting from a classical reconstruction algorithm, an LLM-driven multi-agent search modifies only the components that the user has opened for editing, without prescribing the mechanism to be discovered. Candidates that violate the physical constraints of the task are rejected outright, and the remaining ones are scored by execution. The output is neither an agent system nor a neural network, but a standalone white-box algorithm that requires no agent or neural network at inference time. For interpolation without added noise, the search discovered a residual-gated, phase-aligned dip-consistency projection; Evo-POCS improves the SNR over classic POCS by 3.49 dB on average across missing ratios from 30% to 70%. For simultaneous interpolation and denoising, it discovered a reliability-grouped singular-value shrinkage; Evo-MSSA improves the average reconstruction SNR by more than 7 dB over classic MSSA and by more than 3 dB over a stronger rank-reduction baseline. Both operators retain their gains on data not used during the search. To the best of our knowledge, this is the first study to formulate the design of a seismic reconstruction operator as a constrained, LLM-driven program evolution task. Agentic algorithm evolution can thus complement deep learning in discovering explicit, inspectable, and deployable seismic processing algorithms.

physics.geo-ph↗

arXiv2026

GenRouter: Unified Workflow Routing for Agentic Image Generation

The rapid evolution of text-to-image (T2I) generation models has effectively solved the foundational challenge of raw pixel synthesis, shifting the community's focus toward fulfilling increasingly intricate user requests. While recent agentic image generation workflows enhance static inference with advanced capabilities like external knowledge retrieval and iterative reasoning, they mostly operate in isolated silos with fixed ``one-size-fits-all" topologies. This inevitably leads to severe compute-mismatch, where simple queries are forced through computationally heavy pipelines. To bridge this gap, we present GenRouter, the first unified workflow routing framework for agentic image generation. We first formulate GenCanvas, standardizing diverse agentic pipelines into a universal set of foundational primitives and executable templates. Operating over this unified space, GenRouter adaptively routes heterogeneous prompts to their optimal workflows via (i) demand profiling, (ii) experience matching, and (iii) Pareto filtering. Extensive experiments across diverse benchmarks demonstrate that GenRouter achieves superior visual alignment while reducing execution costs by over 95% and latency by 65% compared to heavyweight static pipelines. Furthermore, the system continuously self-evolves via accumulated experience, enabling robust zero-shot generalization that boosts performance and halves computational overhead.

cs.CV↗

arXiv2026

OpenCodeReview: Determinism over Non-Determinism for Cost-Effective Agent-Based Code Review

LLM-based code review agents promise scalable, always-on review, yet current systems suffer from two intertwined weaknesses: (1) non-determinism--unbounded tool use makes review outcomes unstable, and (2) context locality--the reviewer's access remains bounded to the diff, capping discoverable issue depth. Both give rise to three challenges: misaligned context retrieval, a coherence-efficiency trade-off in multi-file pull requests, and hallucinated comments that erode trust. To address these, we introduce OpenCodeReview, built on deterministic engineering for uncertain agents: rather than granting maximal freedom, we inject determinism at three deliberate pipeline points. Rule-Guided Dispatch uses a multi-layer rule system to deterministically select files and review criteria, eliminating variability of agent-driven triage. Grounded File Review replaces free-form exploration with a curated tool set exposed through a ReAct loop, while file-level parallel SubAgents balance context coherence against efficiency and recover cross-file dependencies on demand. Independent Reflection introduces a falsification-first filter under an asymmetric information boundary--the reflector sees only the diff, not the agent's tool-augmented exploration--removing hallucinated comments without self-reinforcing bias, improving precision while preserving recall. On AACR-Bench (200 real-world PRs, 10 languages, 1,505 expert-verified comments), OpenCodeReview outperforms mainstream coding agents (e.g., Claude Code and Codex) across six LLM backends, achieving up to 2.17x higher SEM-F1 (25.10% vs. 11.57%) while consuming 5-15x fewer tokens. We open-source OpenCodeReview at https://github.com/alibaba/open-code-review.

cs.SE↗

arXiv2026

MC-Risk: Multi-Component Risk Fields for Risk Identification and Motion Planning

We present MC-Risk, a planner-aligned, multi-component risk field on a bird's-eye-view grid that yields early, calibrated, and class-aware risk localization. MC-Risk linearly composes three interpretable modules: (i) a motorized-agent field that fuses a black-box multimodal trajectory predictor with an analytic Gaussian-torus construction whose lateral width grows with speed/curvature and whose height attenuates with look-ahead; (ii) a VRU risk field that replaces isotropic pedestrian blobs with a forward-biased anisotropic kernel aligned to heading and speed; and (iii) a road penalty field that exploits full HD-map topology, imposing an off-road penalty and lane-aware risk exposure for same/opposite directions. We conduct, to our knowledge, the first standardized quantitative evaluation of a risk-field formulation on RiskBench's collision subset. MC-Risk attains the best overall risk localization and the earliest hazard indication. Finally, we demonstrate a plug-and-play planning interface by using the field as an MPC cost density, enabling risk-aware trajectory generation without additional training.

cs.RO↗

arXiv2026

PyraVid: Hierarchical Multimodal Memory for Long-Horizon Video Reasoning

Memory has become an increasingly important component of agentic systems, as these systems are expected to reason over long-term experience. However, prior work has largely focused on unimodal memory, leaving multimodal memory relatively underexplored despite its central role in real-world applications. Compared with unimodal settings, multimodal memory introduces additional challenges, including heterogeneous input integration, person-centric information alignment, and evidence aggregation across different granularities. We present PyraVid, a hierarchical multimodal memory framework inspired by Event Segmentation Theory from cognitive science. PyraVid organizes long videos into a coarse-to-fine pyramid structure, enabling structured memory access and effective evidence aggregation. It further supports structure-guided memory expansion with pruning, allowing the retrieval of related events with strong causal connectivity but low semantic similarity while reducing noise. Experiments on multiple long-video understanding benchmarks show that PyraVid consistently improves performance across datasets, model scales, and question types, highlighting the effectiveness of hierarchical multimodal memory for long-horizon reasoning.

cs.MA↗

arXiv2026

RoboEvolve: Co-Evolving Planner-Simulator for Robotic Manipulation with Limited Data

The scalability of robotic manipulation is fundamentally bottlenecked by the scarcity of task-aligned physical interaction data. While vision-language models (VLMs) and video generation models (VGMs) hold promise for autonomous data synthesis, they suffer from semantic-spatial misalignment and physical hallucinations, respectively. To bridge this gap, we introduce RoboEvolve, a novel framework that couples a VLM planner and a VGM simulator into a mutually reinforcing co-evolutionary loop. Operating purely on unlabeled seed images, RoboEvolve leverages a cognitive-inspired dual-phase mechanism: (i) daytime exploration fosters physically grounded behavioral discovery through a semantic-controlled multi-granular reward, and (ii) nighttime consolidation mines "near-miss" failures to stabilize policy optimization. Guided by an autonomous progressive curriculum, the system naturally scales from simple atomic actions to complex tasks. Extensive experiments demonstrate that RoboEvolve (I) achieves superior effectiveness, elevating base planners by 30 absolute points and amplifying simulator success by 48% on average; (II) exhibits extreme data efficiency, surpassing fully supervised baselines with merely 500 unlabeled seeds--a 50x reduction; and (III) demonstrates robust continual learning without catastrophic forgetting.

cs.RO↗

arXiv2026

GenAI-Driven Approach to RISC-V Supply Chain Exploration

This paper presents an LLM-empowered workflow for RISC-V supply chain analysis, integrating Vision-Language Models (VLMs) and Model-Driven Engineering (MDE) to enable comprehensive, multimodal data-driven insights. The proposed approach addresses the challenges of heterogeneous and unstructured supply chain data by leveraging LLMs for textual understanding and VLMs for extracting information from visual artifacts such as diagrams, tables, and scanned documents. These models collaboratively identify key entities and relationships, which are then organized into a knowledge graph representing supply chain components and their interdependencies. For analytical reasoning, the workflow incorporates MDE techniques and constraint-based modeling to enable formal validation of dependencies, detection of bottlenecks, and assessment of risks. The synergy between LLM- and VLM-based semantic understanding and MDE-based formal analysis supports both exploratory and systematic evaluation of supply chain resilience. A human-in-the-loop mechanism further enables interactive querying and expert validation. The approach is evaluated in RISC-V ecosystem scenarios, demonstrating its effectiveness in generating actionable insights, enhancing transparency, and supporting decision-making in complex semiconductor supply chains.

cs.AR↗

arXiv2026

SegRGB-X: General RGB-X Semantic Segmentation Model

Semantic segmentation across arbitrary sensor modalities faces significant challenges due to diverse sensor characteristics, and the traditional configurations for this task result in redundant development efforts. We address these challenges by introducing a universal arbitrary-modal semantic segmentation framework that unifies segmentation across multiple modalities. Our approach features three key innovations: (1) the Modality-aware CLIP (MA-CLIP), which provides modality-specific scene understanding guidance through LoRA fine-tuning; (2) Modality-aligned Embeddings for capturing fine-grained features; and (3) the Domain-specific Refinement Module (DSRM) for dynamic feature adjustment. Evaluated on five diverse datasets with different complementary modalities (event, thermal, depth, polarization, and light field), our model surpasses specialized multi-modal methods and achieves state-of-the-art performance with a mIoU of 65.03%. The codes will be released upon acceptance.

cs.CV↗

arXiv2026

AgentRob: From Virtual Forum Agents to Hijacked Physical Robots

Large Language Model (LLM)-powered autonomous agents have demonstrated significant capabilities in virtual environments, yet their integration with the physical world remains narrowly confined to direct control interfaces. We present AgentRob, a framework that bridges online community forums, LLM-powered agents, and physical robots through the Model Context Protocol (MCP). AgentRob enables a novel paradigm where autonomous agents participate in online forums--reading posts, extracting natural language commands, dispatching physical robot actions, and reporting results back to the community. The system comprises three layers: a Forum Layer providing asynchronous, persistent, multi-agent interaction; an Agent Layer with forum agents that poll for @mention-targeted commands; and a Robot Layer with VLM-driven controllers and Unitree Go2/G1 hardware that translate commands into robot primitives via iterative tool calling. The framework supports multiple concurrent agents with distinct identities and physical embodiments coexisting in the same forum, establishing the feasibility of forum-mediated multi-agent robot orchestration.

cs.RO↗

arXiv2025

Distributed MIMO Positioning: Fundamental Limit Analysis and User Tracking Framework Design

This paper presents a comprehensive study on the 3D positioning capabilities in distributed multiple-input multiple-output (MIMO) systems. Unlike previous studies that mainly rely on idealized isotropic antenna models, we adopt a polarimetric model that takes advantage of effective aperture distribution functions to characterize realistic antenna patterns, placements, and polarization effects. Based on this model, we analyze the fundamental limits of UE positioning using the Fisher information matrix (FIM) and the position error bound (PEB). The FIM is shown to be expressed as a weighted sum of the information contributions from individual access point (AP)-UE pairs, with each contribution interpreted geometrically across distance, azimuth, and elevation dimensions. The impact of the UE tilt and the spatial distribution of APs on the PEBs is further analyzed. As a further advancement, we propose a complete positioning framework from a UE tracking perspective. By integrating a global probability hypothesis density filter and a PEB-aware AP management strategy, the framework enables accurate tracking while optimizing AP scheduling. Finally, we present a distributed MIMO channel measurement campaign to validate the proposed framework. The results demonstrate a centimeter-level tracking accuracy. In addition, the PEB-aware AP management strategy is shown to maintain robust tracking performance while significantly reducing the number of concurrently active APs, thus lowering the overall system overhead.

eess.SP↗

arXiv2025

3D Cooperative User Tracking for Distributed Integrated Sensing and Communication

As integrated sensing and communication (ISAC) becomes an integral part of 6G networks, distributed ISAC (DISAC) is expected to enhance both sensing and communication performance through its decentralized architecture. This paper presents a complete framework to address the challenge of cooperative user tracking in DISAC systems. By incorporating a global probability hypothesis density (PHD) filter and a field-of-view-aware access point (AP) management strategy, the framework enables accurate user tracking using radio signals while optimizing AP scheduling. In addition, a real-world distributed MIMO channel measurement campaign is performed to evaluate the effectiveness of the framework. The results demonstrate that a centimeter-level root mean-square trajectory error can be achieved. Furthermore, the results show that it is not necessary to keep APs active at all times to maintain high tracking accuracy, indicating the need for robust and efficient AP management. These findings provide valuable insights into practical deployments and the further development of cooperative user tracking techniques in DISAC systems.

eess.SP↗

arXiv2025

Experimental Analysis of Multipath Characteristics in Indoor Distributed Massive MIMO Channels

Distributed multiple-input multiple-output (MIMO), also known as cell-free massive MIMO, has emerged as a promising technology for sixth-generation (6G) wireless networks. This letter introduces an indoor channel measurement campaign designed to explore the behavior of multipath components (MPCs) in distributed MIMO channels. Fully coherent channels were measured between eight distributed uniform planar arrays (128 elements in total) and a 12-meter user equipment route. A method is introduced to determine the order (single- or multi-bounce) of MPC interaction by leveraging map information and MPC parameters. In addition, a Kalman filter-based framework is used for identifying the MPC interaction mechanisms (reflection or scattering/diffraction/mixed). Finally, a comprehensive MPC-level characterization is performed based on the measured channels, including the significance of single-bounce MPCs, common single-bounce MPCs, spherical wavefront features, birth-and-death processes of the MPCs, and spatial distribution of reflections. The findings serve as a valuable reference for understanding the propagation behavior of MPCs, which is necessary for future modeling of indoor distributed MIMO channels.

eess.SP↗

arXiv2025

BuildSTG: A Multi-building Energy Load Forecasting Method using Spatio-Temporal Graph Neural Network

Due to the extensive availability of operation data, data-driven methods show strong capabilities in predicting building energy loads. Buildings with similar features often share energy patterns, reflected by spatial dependencies in their operational data, which conventional prediction methods struggle to capture. To overcome this, we propose a multi-building prediction approach using spatio-temporal graph neural networks, comprising graph representation, graph learning, and interpretation. First, a graph is built based on building characteristics and environmental factors. Next, a multi-level graph convolutional architecture with attention is developed for energy prediction. Lastly, a method interpreting the optimized graph structure is introduced. Experiments on the Building Data Genome Project 2 dataset confirm superior performance over baselines such as XGBoost, SVR, FCNN, GRU, and Naive, highlighting the method's robustness, generalization, and interpretability in capturing meaningful building similarities and spatial relationships.

cs.LG↗

arXiv2025

Advancing high-fidelity 3D and Texture Generation with 2.5D latents

Despite the availability of large-scale 3D datasets and advancements in 3D generative models, the complexity and uneven quality of 3D geometry and texture data continue to hinder the performance of 3D generation techniques. In most existing approaches, 3D geometry and texture are generated in separate stages using different models and non-unified representations, frequently leading to unsatisfactory coherence between geometry and texture. To address these challenges, we propose a novel framework for joint generation of 3D geometry and texture. Specifically, we focus in generate a versatile 2.5D representations that can be seamlessly transformed between 2D and 3D. Our approach begins by integrating multiview RGB, normal, and coordinate images into a unified representation, termed as 2.5D latents. Next, we adapt pre-trained 2D foundation models for high-fidelity 2.5D generation, utilizing both text and image conditions. Finally, we introduce a lightweight 2.5D-to-3D refiner-decoder framework that efficiently generates detailed 3D representations from 2.5D images. Extensive experiments demonstrate that our model not only excels in generating high-quality 3D objects with coherent structure and color from text and image inputs but also significantly outperforms existing methods in geometry-conditioned texture generation.

cs.CV↗

arXiv2025

Sparse Arrays Enable Near-Field Constant-Distance Focusing with Reduced Focal Shift

In near-field beam focusing for finite-sized arrays, focal shift is a non-negligible issue. The actual focal point often appears closer to the array than the predefined focal distance, significantly degrading the focusing performance of finite aperture arrays. Moreover, when the focus point is scanned across different locations, the degradation becomes even more pronounced, leading not only to positional deviation but also to substantial energy loss. To address this issue, we revisit the problem from the perspective of communication degrees of freedom. We demonstrate that a properly designed sparse array with optimized element spacing can effectively mitigate focal shift while enabling stable control of the focusing height during beam scanning. Simulation results based on dipole antennas with different polarizations and patch antennas validate our findings. Notably, with optimized inter-element distances, the energy distribution across focal points becomes nearly uniform, and highly accurate focusing positions are achieved.

eess.SP↗

arXiv2025

Interacting Object-Enabled Clustering and Characterization of Distributed MIMO Channels

Distributed multiple-input multiple-output (MIMO), also known as cell-free massive MIMO, emerges as a promising technology for sixth-generation (6G) systems to support uniform coverage and reliable communication. For the design and optimization of such systems, measurement-based investigations of real-world distributed MIMO channels are essential. In this paper, we present an indoor channel measurement campaign, featuring eight distributed antenna arrays with 128 elements in total. Multi-link channels are measured at 50 positions along a 12-meter user route. A clustering algorithm enabled by interacting objects is proposed to identify clusters in the measured channels. The algorithm jointly clusters the multipath components for all links, effectively capturing the dynamic contributions of common clusters to different links. In addition, a Kalman filter-based tracking framework is introduced for cluster prediction, tracking, and updating along the user movement. Using the clustering and tracking results, cluster-level characterization of the measured channels is performed. First, the number of clusters and their visibility at both link ends are analyzed. Next, a maximum-likelihood estimator is utilized to determine the entire cluster visibility region length. Finally, key cluster-level properties, including the common cluster ratio, cluster power, shadowing, spread, among others, are statistically investigated. The results provide valuable insights into cluster behavior in typical multi-link channels, necessary for accurate modeling of distributed MIMO channels.

eess.SP↗

arXiv2025

Kiss3DGen: Repurposing Image Diffusion Models for 3D Asset Generation

Diffusion models have achieved great success in generating 2D images. However, the quality and generalizability of 3D content generation remain limited. State-of-the-art methods often require large-scale 3D assets for training, which are challenging to collect. In this work, we introduce Kiss3DGen (Keep It Simple and Straightforward in 3D Generation), an efficient framework for generating, editing, and enhancing 3D objects by repurposing a well-trained 2D image diffusion model for 3D generation. Specifically, we fine-tune a diffusion model to generate ''3D Bundle Image'', a tiled representation composed of multi-view images and their corresponding normal maps. The normal maps are then used to reconstruct a 3D mesh, and the multi-view images provide texture mapping, resulting in a complete 3D model. This simple method effectively transforms the 3D generation problem into a 2D image generation task, maximizing the utilization of knowledge in pretrained diffusion models. Furthermore, we demonstrate that our Kiss3DGen model is compatible with various diffusion model techniques, enabling advanced features such as 3D editing, mesh and texture enhancement, etc. Through extensive experiments, we demonstrate the effectiveness of our approach, showcasing its ability to produce high-quality 3D models efficiently.

cs.GR↗