arXiv subjects

Pengyuan Wu

Publications and source records attributed to Pengyuan Wu.

4 recordsLinked to original sources

arXiv2026

WOLF: World Model Guided LiDAR Exploration with Predictive Frontiers

LiDAR-based unmanned aerial vehicle (UAV) exploration builds maps by continually selecting where to observe next. However, decisions based on the measured map provide limited foresight into spatial continuations behind occlusions, leaving potentially informative directions unrecognized. We present WOLF, a world-model-guided framework that predicts future observations to enhance autonomous exploration. In the training stage, a recurrent world model learns observation dynamics from exploration trajectories, with recurrent memory retaining the spatial context needed to interpret partial observations across successive views. Building on this context, the model combines observation history with candidate motions during exploration to predict local occupancy and visibility. To guide further sensing, a predictive frontier generation mechanism then aligns and fuses these predictions using confidence, branch agreement, and observation quality to identify promising regions. The resulting predictive frontiers join measured ones to guide geometric viewpoint selection and trajectory generation, while new scans update subsequent predictions. In simulations, our method reduces mean terminal time by 10.9% relative to EPIC in Garage at comparable coverage and increases mean coverage from 42.12% to 98.35% in Tunnel. Real-world experiments further demonstrate onboard deployment of the learned model for online inference during physical flight.

cs.RO↗

arXiv2026

Agent-authored deposition recipes for X-ray multilayer mirrors: schema-bound LLM control of a magnetron sputtering system with reflectivity-verified outcomes

Periodic multilayer mirrors for X-rays require a layer period controlled to about 0.1 nm, set by deposition rates held in the laboratory's own notebook. A supervising model session gave a large language model agent a target multilayer specification and calibration constants derived from a laboratory archive. The agent authored recipes in the magnetron sputtering system's native recipe language and submitted them through a Model Context Protocol bridge of 17 tools in three tiers, with enforcement in the control application. The target was a 30-bilayer Ru/C multilayer with a period of 6.85 nm, a pair chosen because each period is fitted as two layers. The agent could read and write only its own working project, so it could alter neither the calibration record nor the measurements that scored it. We scored mirrors by X-ray reflectivity and fixed the 3% acceptance threshold on relative period difference before deposition. Across four scored depositions, the relative period difference fell from 20% on the calibration as given to 2.26% after one measured-feedback iteration and 0.79% after a second. We identified sixteen failure modes of the agent-facing interface. The result is bounded to the period, pair, and chamber observed.

cond-mat.mtrl-sci↗

arXiv2026

Closed-Loop Action Chunks with Dynamic Corrections for Training-Free Diffusion Policy

Diffusion-based policies have achieved remarkable results in robotic manipulation but often struggle to adapt rapidly in dynamic scenarios, leading to delayed responses or task failures. We present DCDP, a Dynamic Closed-Loop Diffusion Policy framework that integrates chunk-based action generation with real-time correction. DCDP integrates a self-supervised dynamic feature encoder, cross-attention fusion, and an asymmetric action encoder-decoder to inject environmental dynamics before action execution, achieving real-time closed-loop action correction and enhancing the system's adaptability in dynamic scenarios. In dynamic PushT simulations, DCDP improves adaptability by 19\% without retraining while requiring only 5\% additional computation. Its modular design enables plug-and-play integration, achieving both temporal coherence and real-time responsiveness in dynamic robotic scenarios, including real-world manipulation tasks. The project page is at: https://github.com/wupengyuan/dcdp

cs.RO↗

arXiv2025

Cross from Left to Right Brain: Adaptive Text Dreamer for Vision-and-Language Navigation

Vision-and-Language Navigation (VLN) requires the agent to navigate by following natural instructions under partial observability, making it difficult to align perception with language. Recent methods mitigate this by imagining future scenes, yet they rely on vision-based synthesis, leading to high computational cost and redundant details. To this end, we propose to adaptively imagine key environmental semantics via \textit{language} form, enabling a more reliable and efficient strategy. Specifically, we introduce a novel Adaptive Text Dreamer (ATD), a dual-branch self-guided imagination policy built upon a large language model (LLM). ATD is designed with a human-like left-right brain architecture, where the left brain focuses on logical integration, and the right brain is responsible for imaginative prediction of future scenes. To achieve this, we fine-tune only the Q-former within both brains to efficiently activate domain-specific knowledge in the LLM, enabling dynamic updates of logical reasoning and imagination during navigation. Furthermore, we introduce a cross-interaction mechanism to regularize the imagined outputs and inject them into a navigation expert module, allowing ATD to jointly exploit both the reasoning capacity of the LLM and the expertise of the navigation model. We conduct extensive experiments on the R2R benchmark, where ATD achieves state-of-the-art performance with fewer parameters. The code is \href{https://github.com/zhangpingrui/Adaptive-Text-Dreamer}{here}.

cs.CV↗