[1]李泚泚,田国会,路飞,等.面向服务机器人的物品知识自主构建方法[J].东南大学学报(自然科学版),2020,50(2):395-401.[doi:10.3969/j.issn.1001-0505.2020.02.025]
 Li Cici,Tian Guohui,Lu Fei,et al.Automatic construction method for object knowledge on service robots[J].Journal of Southeast University (Natural Science Edition),2020,50(2):395-401.[doi:10.3969/j.issn.1001-0505.2020.02.025]
点击复制

面向服务机器人的物品知识自主构建方法()
分享到:

《东南大学学报(自然科学版)》[ISSN:1001-0505/CN:32-1178/N]

卷:
50
期数:
2020年第2期
页码:
395-401
栏目:
自动化
出版日期:
2020-03-20

文章信息/Info

Title:
Automatic construction method for object knowledge on service robots
作者:
李泚泚田国会路飞张森彦
山东大学控制科学与工程学院, 济南 250061
Author(s):
Li Cici Tian Guohui Lu Fei Zhang Senyan
School of Control Science and Engineering, Shandong University, Jinan 250061, China
关键词:
服务机器人 物品知识 知识构建 本体技术 智能操作
Keywords:
service robot object knowledge knowledge construction ontology technology intelligent operation
分类号:
TP242
DOI:
10.3969/j.issn.1001-0505.2020.02.025
摘要:
为了提高家庭服务机器人操作物品时的智能性,提出了一种层次化的物品知识自动构建方法.首先,在类层级上,利用本体技术构建基于视觉、类别、物理、功能、状态、位置等多属性的物品知识表征模板,以便于以统一的、机器可读的、可共享的形式表征物品知识;其次,在实例层级上,提出基于视觉属性的物品实例获取方法,并基于物品实例知识表征机制,完成物品实例知识的自主构建;最后,研究了物品知识的自主推理方法,以获取隐含知识.实验结果表明:物品知识构建方法可将获取的源数据自主转换成本体知识,根据所提出的方法成功构建了包含20类物品、568个实例、117条规则在内的物品本体模型;所设计的SWRL规则可自主推理本体实例缺失的属性且所需时间少于1.2 s;物品本体知识不仅可为物品操作提供先验知识,而且在目标物品缺失时可提供替代物品.
Abstract:
To make the home service robot operate objects more intelligently, an automatic construction method for hierarchical object knowledge was proposed. First, in the class hierarchy, an object knowledge representation template with multiple attributes, such as visual, category, physical, affordance, state, and location, was constructed, so that the object knowledge was represented in a unified, machine-readable and shared form. Secondly, in the instance hierarchy, a method for recognizing object instances with their visual attributes was proposed. Subsequently, with the object instance knowledge representation mechanism, the object instance knowledge was constructed. Finally, to obtain the implicit knowledge, an automatic reasoning method for the object knowledge was studied. The experimental results indicate that the construction method for the object knowledge can automatically convert the source data to ontology knowledge. According to the proposed method, the object ontology model including 20 kinds of objects, 568 instances and 117 rules is successfully constructed. The designed SWRL rules can reason about the missing attributes of instances in the object ontology, and the time required is less than 1.2 s. The object ontology knowledge provides the robot with prior knowledge when operating on the object, and provides the robot with a substitute when the target object is missing.

参考文献/References:

[1] ten Pas A, Platt R. Localizing handle-like grasp affordances in 3D point clouds[M]//Experimental Robotics. Cham: Springer International Publishing, 2016, 109: 623-638. DOI:10.1007/978-3-319-23778-7_41.
[2] 李国栋, 田国会, 薛英花. 基于QR Code技术的家庭服务机器人视觉伺服抓取操作研究[J]. 东南大学学报(自然科学版), 2010, 40(S1): 30-36.
  Li G D, Tian G H, Xue Y H. Research on QR code-based visual servo handling of room service robot[J]. Journal of Southeast University(Natural Science Edition), 2010, 40(S1): 30-36.(in Chinese)
[3] Feix T, Bullock I M, Dollar A M. Analysis of human grasping behavior: Object characteristics and grasp type[J]. IEEE Transactions on Haptics, 2014, 7(3): 311-323. DOI:10.1109/toh.2014.2326871.
[4] Farhadi A, Endres I, Hoiem D, et al. Describing objects by their attributes[C]//2009 IEEE Conference on Computer Vision and Pattern Recognition. Miami, FL, USA, 2009: 1778-1785. DOI:10.1109/cvpr.2009.5206772.
[5] Zhu Y K,Fathi A, Li F F. Reasoning about object affordances in a knowledge base representation[M]//Computer Vision- ECCV 2014. Cham: Springer International Publishing, 2014, 8690: 408-424. DOI:10.1007/978-3-319-10605-2_27.
[6] Waibel M, Beetz M, Civera J, et al. RoboEarth[J]. IEEE Robotics & Aut omation Magazine, 2011, 18(2): 69-82. DOI:10.1109/mra.2011.941632.
[7] 张营, 田国会, 张森彦, 等. 家庭智能空间下多领域知识的共享与重用方法[J]. 机器人, 2019, 41(4): 507-518. DOI:10.13973/j.cnki.robot.180494.
Zhang Y, Tian G H, Zhang S Y, et al. Multi-domain knowledge sharing and reuse in home intelligent space[J]. Robot, 2019, 41(4): 507-518. DOI:10.13973/j.cnki.robot.180494. (in Chinese)
[8] 田国会, 王晓静, 张营. 一种家庭服务机器人的环境语义认知机制[J]. 华中科技大学学报(自然科学版), 2018, 46(12): 18-23. DOI:10.13245/j.hust.181204.
Tian G H, Wang X J, Zhang Y. An environmental semantic cognition mechanism of home service robots[J]. Journal of Huazhong University of Science and Technology(Natural Science Edition), 2018, 46(12): 18-23. DOI:10.13245/j.hust.181204. (in Chinese)
[9] Antoniou G, Harmelen F. 语义网基础教程[M]. 陈小平, 等译. 北京: 机械工业出版社, 2008: 85-98.
[10] Redmon J, Farhadi A. YOLO9000: Better, faster, stronger[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition(CVPR). Honolulu, HI, USA, 2017: 6517-6525. DOI:10.1109/cvpr.2017.690.
[11] Danaci E G, Ikizler-Cinbis N. Low-level features for visual attribute recognition: An evaluation[J]. Pattern Recognition Letters, 2016, 84: 185-191. DOI:10.1016/j.patrec.2016.09.015.
[12] Scheirer W J, Kumar N, Belhumeur P N, et al. Multi-attribute spaces: Calibration for attribute fusion and similarity search[C]//2012 IEEE Conference on Computer Vision and Pattern Recognition. Providence, RI, USA, 2012: 2933-2940. DOI:10.1109/cvpr.2012.6248021.
[13] 路飞, 田国会, 李擎. 智能空间环境下基于本体的机器人服务自主认知及规划[J]. 机器人, 2017, 39(4): 423-430.
  Lu F, Tian G H, Li Q. Autonomous cognition and planning of robot service based on ontology in intelligent space environment [J]. Robot, 2017, 39(4): 423-430.(in Chinese)

备注/Memo

备注/Memo:
收稿日期: 2019-08-07.
作者简介: 李泚泚(1990—),女,博士生;田国会(联系人),男,博士,教授,博士生导师,g.h.tian@sdu.edu.cn.
基金项目: 国家自然科学基金联合基金重点资助项目(U1813215)、国家自然科学基金资助项目(61773239, 61973187).
引用本文: 李泚泚,田国会,路飞,等.面向服务机器人的物品知识自主构建方法[J].东南大学学报(自然科学版),2020,50(2):395-401. DOI:10.3969/j.issn.1001-0505.2020.02.025.
更新日期/Last Update: 2020-03-20