大语言模型在机器人任务规划中的应用研究综述  

Survey of the application of large language model in robot task planning

作　　者：郭庭航 李晓寒[1] 殷芊涵 季新 秦转萍 路光达[1,2] GUO Tinghang;LI Xiaohan;YIN Qianhan;JI Xin;QIN Zhuanping;LU Guangda(School of Automation and Electrical Engineering,Tianjin University of Technology and Education,Tianjin 300222,China;Tianjin Key Laboratory of Information Sensing&Intelligent Control,Tianjin University of Technology and Education,Tianjin 300222,China)

机构地区：[1]天津职业技术师范大学自动化与电气工程学院,天津300222 [2]天津职业技术师范大学天津市信息传感与智能控制重点实验室,天津300222

出　　处：《天津职业技术师范大学学报》2025年第1期12-18,共7页Journal of Tianjin University of Technology and Education

基　　金：天津市“揭榜挂帅”科技计划项目(2023JB02);天津市教委科研计划项目(2021KJ011).

摘　　要：机器人任务规划是指根据任务需求、环境条件以及机器人的能力,为机器人制定一系列详细的行动方案,其在医疗健康、制造业、家庭服务、教育和空间探索等领域广泛应用。大语言模型(LLM),以其强大的自然语言处理能力、深度理解并生成文本的能力以及广泛的知识面等特点,为机器人任务规划带来了适应性和灵活性等优势,成为未来的发展趋势之一。文章通过概述机器人任务规划的发展历程,并从环境感知、行为分解和运动规划3个层面论述了以Transformer为核心架构的典型大语言模型及其他基础模型在机器人任务规划领域的研究与应用,总结了大语言模型应用在机器人任务规划中的优势及挑战,以期为后续研究提供参考。Robot task planning involves devising a series of detailed action plans for a robot based on its capabilities,task requirements,and environmental conditions,and has a wide range of applications in the fields of healthcare,manufacturing,home services,education,and space exploration.Large language models,characterized by their powerful natural language processing capabilities,deep understanding and generation of text,as well as extensive knowledge,offer advantages of adaptability and flexibility for robot task planning,positioning them as a trending development for the future.This paper reviews the development of robot task planning and discusses the research and application of the large language model with the Transformer as the core architecture and other basic models in the field of robot task planning from three levels,namely,environment perception,behavior decomposition and motion planning.Finally,the advantages and challenges of applying the large language model to robot task planning are summarized to provide reference for research in related fields.

关 键 词：大语言模型 任务分解 高级规划 逻辑推理 

分 类 号：TM242[一般工业技术—材料科学与工程]