通信学报2025，Vol.46Issue(12)：1-13,13.DOI:10.11959/j.issn.1000-436x.2025182
具身智能驱动的多AMR通信与控制协同优化研究
Embodied intelligence-driven collaborative optimization of communication and control in multi-AMR systems
摘要
Abstract
To meet communication-control co-design requirements of multiple autonomous mobile robots(AMR)per-forming high-frequency,multi-stage tasks in 6G industrial scenarios,an embodied intelligence system was developed to fuse communication enhancement and trajectory-aware perception.The system was composed of three layers,including embodied interaction,communication collaboration,and intelligent control,which were responsible for local sensing,downlink enhancement and scheduling,and centralized policy learning with online updates,respectively.To overcome the limitations of traditional separated approaches that rely on prior maps and separate communication from control,non-orthogonal multiple access(NOMA)was combined with multi-antenna beamforming.Meanwhile,task scheduling,path planning,and communication resource allocation were jointly formulated as a unified task-control-communication prob-lem.To handle the high-dimensional coupling and environmental dynamics,a hierarchical deep reinforcement learning(DRL)framework with parallel sampling,centralized training and distributed execution was employed.Simulation re-sults show that the proposed architecture rapidly constructs accurate accessibility and channel estimation maps without prior knowledge,and maintains high task completion rates and communication performance across diverse obstacle set-tings,demonstrating its robustness and efficiency.关键词
具身智能/自主移动机器人/非正交多址接入/深度强化学习/轨迹控制Key words
embodied intelligence/AMR/NOMA/DRL/trajectory control分类
信息技术与安全科学引用本文复制引用
罗如瑜,高天润,王嘉诚,田辉,张平..具身智能驱动的多AMR通信与控制协同优化研究[J].通信学报,2025,46(12):1-13,13.基金项目
国家自然科学基金资助项目(No.62071068) （No.62071068）
北京邮电大学博士生创新基金资助项目(No.CX2023144)The National Natural Science Foundation of China(No.62071068),BUPT Excellent Ph.D.Students Foundation(No.CX2023144) （No.CX2023144）
