记忆神经网络在机器人导航领域的应用与研究进展  

Research progress and application of memory neural network in robot navigation

作　　者：王作为 徐征 张汝波[5] 洪才森 王殊 WANG Zuowei;XU Zheng;ZHANG Rubo;HONG Caisen;WANG Shu(School of Computer Science and Technology,Tianjin Polytechnic University,Tianjin 300387,China;College of Mechanical Engineering Post-doctoral Research Station,Tianjin Polytechnic University,Tianjin 300387,China;DongHexin Technology Co.,Ltd.,Tianjin 300350,China;College of Automobile and Transportation,Tianjin University of Technology and Education,Tianjin 300222,China;College of Mechanical and Electrical Engineering,Dalian Minzu University,Dalian 116600,China)

机构地区：[1]天津工业大学计算机科学与技术学院,天津300387 [2]天津工业大学机械工程学院博士后工作站,天津300387 [3]天津动核芯科技有限公司,天津300350 [4]天津职业技术师范大学汽车与交通学院,天津300222 [5]大连民族大学机电工程学院,辽宁大连116600

出　　处：《智能系统学报》2020年第5期835-846,共12页CAAI Transactions on Intelligent Systems

基　　金：国家自然科学基金面上项目(61972456);天津市教委科研计划项目(2019KJ018);天津工业大学学位与研究生教育改革项目(Y20180104).

摘　　要：记忆神经网络非常适合解决时间序列决策问题,将其用于机器人导航领域是非常有前景的新兴研究领域。本文主要讨论记忆神经网络在机器人导航领域的研究进展。给出几种基本记忆神经网络结合导航任务的工作机理,总结了不同模型的优缺点;对记忆神经网络在导航领域的研究进展进行简要综述;进一步介绍导航验证环境的发展;最后梳理了记忆神经网络在导航问题所面临的复杂性挑战,并预测了记忆神经网络在导航领域未来的发展方向。Memory networks are a relatively new class of models designed to alleviate the problem of learning long-term dependencies in sequential data,by providing an explicit memory representation for each token in the sequence,and they can be used for learning navigation policies in an unstructured terrain,which is a complex task.Memory neural networks are highly suitable for solving time series decision-making problems,and their application in robot navigation is a very promising and emerging research field.The research progress of memory neural networks in the field of robot navigation is primarily discussed in this paper.First,the working mechanism of several basic memory neural networks used for robot navigationis introduced,and the advantages and disadvantages of different models are summarized.Then,the research progress of memory neural network in navigation field is briefly reviewed,and the development of navigation verification environment is discussed.Finally,the complex challenges faced by memory neural networks in navigation are summarized,and the future development of memory neural networks in navigation field is predicted.

关 键 词：记忆神经网络 机器人导航 深度强化学习 可微神经计算机 可微神经字典 深度学习 强化学习 记忆网络 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]