Vision-Language-Action Models Meet World Models: Embodied Agentic AI for Low-Altitude Wireless Networks

2026-06-09

AI Summary

This work addresses three key challenges in deploying large models within low-altitude wireless networks: constrained embodied action mapping, inadequate environmental modeling, and the absence of closed-loop optimization. To overcome these limitations, the authors propose an embodied intelligence framework for unmanned aerial vehicles that integrates a vision-language-action (VLA) model with a world model (WM). This architecture enables, for the first time, end-to-end decision-making from multimodal perception to continuous control. By incorporating memory and reflection mechanisms, it establishes an adaptive closed-loop optimization paradigm—“decide, act, evaluate, update”—that supports environment prediction, policy validation, and dynamic evolution. Evaluated in complex, dynamic low-altitude environments, the system demonstrates significantly enhanced robustness, predictability, and capability for continual autonomous adaptation.

Abstract

Low-Altitude Wireless Networks (LAWNs), composed of Unmanned Aerial Vehicles (UAVs) and other aerial platforms, provide integrated perception, communication, and computation services in low-altitude airspace. However, deploying large generative models in this domain faces three major challenges: 1) Limited embodied action mapping; 2) Inadequate physical environment modeling; 3) Insufficient closed-loop optimization. To address these challenges, this study proposes an Embodied Agentic UAV framework. Centered on a Vision-Language-Action (VLA) model as the execution core, the framework establishes an end-to-end embodied decision-making pipeline from multimodal environmental perception to continuous control generation. In addition, a World Model (WM) is introduced to capture the coupling between UAV actions and environmental state evolution, thereby supporting environment prediction, policy verification, and dynamic optimization. Furthermore, memory and reflection mechanisms are incorporated to form an adaptive closed-loop optimization paradigm of decision, execution, evaluation, and update, thereby enhancing the system's autonomous decision-making capability and continual evolution ability in complex dynamic environments. Experimental results validate its effectiveness in enabling robust, predictive, and sustainable autonomous control in LAWNs.

Problem

• Embodied AI
• Low-Altitude Wireless Networks
• Vision-Language-Action Models
• World Models
• Autonomous Control

Innovation

• Vision-Language-Action Model
• World Model
• Embodied AI
• Closed-loop Optimization
• Low-Altitude Wireless Networks