<-上一篇/Previous Article 下一篇/Next Article->

[1]马成宇,刘华平,葛泉波.场景感知的分布式多智能体目标搜索方法[J].智能系统学报,2022,17(6):1244-1253.[doi:10.11992/tis.202110012]
　MA Chengyu,LIU Huaping,GE Quanbo.Scene-aware decentralized Monte Carlo Tree Search of target discovery[J].CAAI Transactions on Intelligent Systems,2022,17(6):1244-1253.[doi:10.11992/tis.202110012]

点击复制

场景感知的分布式多智能体目标搜索方法

PDF下载 HTML

《智能系统学报》[ISSN 1673-4785/CN 23-1538/TP] 卷: 17 期数: 2022年第6期页码: 1244-1253 栏目: 吴文俊人工智能科学技术奖论坛出版日期: 2022-11-05

Title:: Scene-aware decentralized Monte Carlo Tree Search of target discovery

作者:: 马成宇¹, 刘华平², 葛泉波^3,4; 1. 南通大学电气工程学院，江苏南通 226019;
2. 清华大学计算机科学与技术系，北京 100084;
3. 同济大学电子与信息工程学院，上海 201804;
4. 南京信息工程大学自动化学院，江苏南京 210044

Author(s):: MA Chengyu¹, LIU Huaping², GE Quanbo^3,4; 1. School of Instrumentation and Electrical Engineering, Nantong University, Nantong 226019, China;
2. Department of Computer Science and Technology, Tsinghua University, Beijing 100084, China;
3. School of Electronics and Information Engineering, Tongji University, Shanghai 201804, China;
4. School of Automation, Nanjing University of Information Science and Technology, Nanjing 210044, China

关键词:: 场景图谱; 分布式; 目标搜索; 蒙特卡洛树搜索; 多目标优化; 动作规划; 多智能体系统; 视觉语义导航

Keywords:: scene graph; decentralization; target search; Monte Carlo Tree Search; multi-objective optimization; action planning; multi-agent system; visual semantic navigation

分类号:: TP391

DOI:: 10.11992/tis.202110012

文献标志码:: 2022-10-08

摘要:: 在视觉语义导航任务中，智能体通过视觉信息，寻找并导航到给定对象类别的目标处。然而，大部分现有的研究都是使用基于学习的框架来完成任务，这些研究在现实世界中应用的训练成本非常高，可移植性很低，并且它们只适用于单智能体，效率低下、容错能力差。为解决上述问题，本文提出一种基于场景感知的分布式多目标优化蒙特卡洛树搜索模型，该模型中多智能体实时在线规划并且不需要预先训练，利用场景感知先验知识结合观测信息实时对环境进行估计，并且利用改进的蒙特卡洛树搜索进行路径规划以此搜索目标。在Mtterport3D数据集中进行的实验表明，该模型在效率方面比单智能体有着显著的提高。

Abstract:: In visual semantic navigation, agents find and navigate to the target of a given object category through visual information. However, the majority of existing studies complete the task using a learning-based framework. These studies have a high training cost in the real world, low portability, and are only suitable for single-agent systems with low efficiency and poor fault tolerance. To address the above issues, this paper suggests a decentralized multi-objective optimization Monte Carlo Tree Search model based on scene awareness. In this model, multi-agent plans online in real time and does not need to train in advance. The improved Monte Carlo Tree Search is used for path planning to search targets, and the environment is estimated in real-time by using scene awareness prior knowledge combined with observation information. Experiments in the Matterport3D dataset demonstrate that the effectiveness of the model is significantly higher than that of a single agent.

参考文献/References:: [1] ANDERSON P, WU Qi, TENEY D, et al. Vision-and-language navigation: interpreting visually-grounded navigation instructions in real environments[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City: IEEE, 2018 : 3674-3683.
[2] DAS A, DATTA S, GKIOXARI G, et al. Embodied question answering[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City: IEEE, 2018 : 1-10.
[3] ZHU Yuke, MOTTAGHI R, KOLVE E, et al. Target-driven visual navigation in indoor scenes using deep reinforcement learning[C]//2017 IEEE International Conference on Robotics and Automation. Singapore: IEEE, 2017: 3357-3364.
[4] CHAPLOT D S, GANDHI D, GUPTA A, et al. Object goal navigation using goal-oriented semantic exploration[EB/OL]. (2020-07-02)[2021-10-13]. https://arxiv.org/abs/2007.00643v2.
[5] LIANG Yiqing, CHEN Boyuan, SONG Shuran. SSCNav: confidence-aware semantic scene completion for visual semantic navigation[C]//2021 IEEE International Conference on Robotics and Automation. Xi’an: IEEE, 2021: 13194-13200.
[6] DRUON R, YOSHIYASU Y, KANEZAKI A, et al. Visual object search by learning spatial context[J]. IEEE robotics and automation letters, 2020, 5(2): 1279–1286.
[7] CAMPARI T, ECCHER P, SERAFINI L, et al. Exploiting scene-specific features for object goal navigation[C]//European Conference on Computer Vision. Cham: Springer, 2020: 406-421.
[8] YE Xin, YANG Yezhou. Efficient robotic object search via HIEM: hierarchical policy learning with intrinsic-extrinsic modeling[J]. IEEE robotics and automation letters, 2021, 6(3): 4425–4432.
[9] MOUSAVIAN A, TOSHEV A, FI?ER M, et al. Visual representations for semantic target driven navigation[C]//2019 International Conference on Robotics and Automation (ICRA). Montreal: IEEE, 2019 : 8846-8852.
[10] WU Qiaoyun, GONG Xiaoxi, XU Kai, et al. Towards target-driven visual navigation in indoor scenes via generative imitation learning[J]. IEEE robotics and automation letters, 2021, 6(1): 175–182.
[11] 王贺彬, 葛泉波, 刘华平, 等. 面向观测融合和吸引因子的多机器人主动SLAM[J]. 智能系统学报, 2021, 16(2): 371–377
WANG Hebin, GE Quanbo, LIU Huaping, et al. Multi-robot active SLAM for observation fusion and attractor[J]. CAAI transactions on intelligent systems, 2021, 16(2): 371–377
[12] LIN Qiuzhen, LIU Songbai, ZHU Qingling, et al. Particle swarm optimization with a balanceable fitness estimation for many-objective optimization problems[J]. IEEE transactions on evolutionary computation, 2018, 22(1): 32–46.
[13] 楼传炜, 葛泉波, 刘华平, 等. 无人机群目标搜索的主动感知方法[J]. 智能系统学报, 2021, 16(3): 575–583
LOU Chuanwei, GE Quanbo, LIU Huaping, et al. Active perception method for UAV group target search[J]. CAAI transactions on intelligent systems, 2021, 16(3): 575–583
[14] 吴莹莹, 丁肇红, 刘华平, 等. 面向环境探测的多智能体自组织目标搜索算法[J]. 智能系统学报, 2020, 15(2): 289–295
WU Yingying, DING Zhaohong, LIU Huaping, et al. Self-organizing target search algorithm of multi-agent system for envi-ronment detection[J]. CAAI transactions on intelligent systems, 2020, 15(2): 289–295
[15] LUO Tianze, SUBAGDJA B, WANG Di, et al. Multi-agent collaborative exploration through graph-based deep reinforcement learning[C]//2019 IEEE International Conference on Agents. Jinan: IEEE, 2019 : 2-7.
[16] HU Jinwen, XIE Lihua, XU Jun, et al. Multi-agent cooperative target search[J]. Sensors (Basel, Switzerland), 2014, 14(6): 9408–9428.
[17] JAIN U, WEIHS L, KOLVE E, et al. Two body problem: collaborative visual task completion[C]//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Long Beach: IEEE, 2019 : 6682-6692.
[18] JAIN U, WEIHS L, KOLVE E, et al. A cordial sync: going beyond marginal policies for multi-agent embodied tasks[M]//Computer Vision–ECCV 2020. Cham: Springer International Publishing, 2020: 471-490.
[19] YANG WEI, WANG XIAOLONG, FARHADI A, et al. Visual semantic navigation using scene priors[EB/OL]. (2018-10-15) [ 2021-10-13]. https://arxiv.org/abs/1810.06543.
[20] QIU Yiding, PAL A, CHRISTENSEN H. Learning hierarchical relationships for object-goal navigation[EB/OL]. (2020-11-18) [2021-10-13]. https://arxiv.org/abs/2003.06749v2.
[21] SILVER D, SCHRITTWIESER J, SIMONYAN K, et al. Mastering the game of Go without human knowledge[J]. Nature, 2017, 550(7676): 354–359.
[22] 孙润稼, 刘玉田. 基于深度学习和蒙特卡洛树搜索的机组恢复在线决策[J]. 电力系统自动化, 2018, 42(14): 40–47
SUN Runjia, LIU Yutian. Online decision-making for generator start-up based on deep learning and Monte Carlo tree search[J]. Automation of electric power systems, 2018, 42(14): 40–47
[23] 邱云飞, 于智龙, 郭羽含, 等. 蒙特卡洛树搜索下的整合多目标可持续闭环供应链网络优化[J]. 计算机集成制造系统, 2022, 28(1): 269–293
QIU Yunfei, YU Zhilong, GUO Yuhan, et al. Integrated multi-objective sustainable closed-loop supply chain network optimization under MCTS[J]. Computer integrated manufacturing systems, 2022, 28(1): 269–293
[24] CHEN Weizhe, LIU Lantao. Pareto Monte Carlo tree search for multi-objective informative planning[EB/OL]. (2021-11-02) [2021-10-13]. https://arxiv.org/abs/2111.01825.
[25] BEST G, CLIFF O M, PATTEN T, et al. Dec-MCTS: Decentralized planning for multi-robot active perception[J]. The international journal of robotics research, 2019, 38(2/3): 316–337.
[26] SUKKAR F, BEST G, YOO C, et al. Multi-robot region-of-interest reconstruction with dec-MCTS[C]//2019 International Conference on Robotics and Automation (ICRA). Montreal: IEEE, 2019 : 9101-9107.
[27] KRISHNA R, ZHU Yuke, GROTH O, et al. Visual genome: connecting language and vision using crowdsourced dense image annotations[J]. International journal of computer vision, 2017, 123(1): 32–73.
[28] CHANG A, DAI A, FUNKHOUSER T, et al. Matterport3D: learning from RGB-D data in indoor environments[C]//2017 International Conference on 3D Vision. Qingdao: IEEE, 2017: 667-676.

相似文献/References:: [1]周方波,赵怀林,刘华平.基于场景图谱的室内移动机器人目标搜索[J].智能系统学报,2022,17(5):1032.[doi:10.11992/tis.202109011]
　ZHOU Fangbo,ZHAO Huailin,LIU Huaping.Indoor mobile robot target search based on the scene graphs[J].CAAI Transactions on Intelligent Systems,2022,17():1032.[doi:10.11992/tis.202109011]
[2]陆升阳,赵怀林,刘华平.场景图谱驱动目标搜索的多智能体强化学习[J].智能系统学报,2023,18(1):207.[doi:10.11992/tis.202111034]
　LU Shengyang,ZHAO Huailin,LIU Huaping.Multi-agent reinforcement learning for scene graph-driven target search[J].CAAI Transactions on Intelligent Systems,2023,18():207.[doi:10.11992/tis.202111034]
[3]崔铁军,李莎莎.基于联系数和系统功能状态的因素联系分布式构建与应用[J].智能系统学报,2023,18(6):1305.[doi:10.11992/tis.202208029]
　CUI Tiejun,LI Shasha.Construction and application of factor connection distribution formula based on connection number and system function status[J].CAAI Transactions on Intelligent Systems,2023,18():1305.[doi:10.11992/tis.202208029]

备注/Memo

收稿日期:2021-10-13。
基金项目:国家自然科学基金项目(U1613212).
作者简介:马成宇，硕士研究生，主要研究方向为多智能体系统;刘华平，副教授，博士生导师，中国人工智能学会理事、中国人工智能学会认知系统与信息处理专业委员会秘书长，主要研究方向为机器人感知、学习与控制、多模态信息融合。主持国家自然科学基金重点项目2项。吴文俊人工智能科学技术奖获得者。发表学术论文300余篇;葛泉波，教授，博士生导师，主要研究方向为工程信息融合方法及应用、人机混合系统智能评估。主持国家自然科学基金青年基金项目1项。发表学术论文100余篇
通讯作者:刘华平.E-mail:hpliu@tsinghua.edu.cn

更新日期/Last Update: 1900-01-01

场景感知的分布式多智能体目标搜索方法 PDF下载HTML

备注/Memo

场景感知的分布式多智能体目标搜索方法

PDF下载 HTML