计算机科学与探索 ›› 2018, Vol. 12 ›› Issue (10): 1622-1632.DOI: 10.3778/j.issn.1673-9418.1709028

• 人工智能与模式识别 • 上一篇    下一篇

室内移动机器人人机交互的语义实现

王恒升1,2,任  晋1+   

  1. 1. 中南大学 机电工程学院,长沙 410083
    2. 中南大学 高性能复杂制造国家重点实验室,长沙 410083
  • 出版日期:2018-10-01 发布日期:2018-10-08

Semantic Realization of Human Robot Interaction for Indoor Environments

WANG Hengsheng1,2,REN Jin1+   

  1. 1. College of Mechanical & Electrical Engineering, Central South University, Changsha 410083, China
    2. State Key Laboratory for High Performance Complex Manufacturing, Central South University, Changsha 410083, China
  • Online:2018-10-01 Published:2018-10-08

摘要: 自然语言与机器人远程人机交互的主要困难是语义理解。基于本体的方法建立人机交互的概念体系,用Prolog逻辑编程语言对概念进行定义,形成概念语义事实库;进一步以这些事实库为基础,定义谓词逻辑关系,形成基于概念的谓词逻辑推理机制,从而实现人与机器人在统一的概念内涵和推理逻辑基础上的人机交互。以人与处于室内的远程机器人通过交互完成导航为背景,建立某建筑物内部地名词的概念本体,用Prolog定义了其概念事实;定义了谓词逻辑关系,从事实库中提取出建筑物内部的语义地图;在此基础上,通过逻辑推理的方法实现了机器人室内导航所需的自定位功能和路径规划功能。该内容作为机器人基于本体概念体系的逻辑推理实例,所有的结果均建立在实际数据之上,所有的谓词逻辑均在SWI-Prolog中调试通过,推理结果满足预期。

关键词: 人机交互, 语义地图, 机器人导航, 逻辑编程

Abstract: It is the semantic understanding that makes the main difficulty for the interaction between the operator and the remote robot through natural languages. This paper establishes the concept system for human robot interaction (HRI) based on ontology, and the concepts are defined with Prolog programming language, which forms the rules to be referred to as the semantic contents of the concepts. Predicates of logic relations of concepts are designed based on the rules, which is the basis of reasoning mechanism based on concepts. Then the human-robot interaction works on the consensus meanings of concepts and logics of reasoning basis. The background of this work is the navigation of an indoor mobile robot remotely instructed by an operator with natural spoken language. An ontology is established for the place names of a building, based on which are the Prolog facts. Predicates of logic relations are designed for the purpose of navigation through reasoning, namely, predicate for extracting the semantic map, predicate for self-  localization of the robot, predicate for path planning. This paper works as a successful case study of realizing logic reasoning of robots through an ontology of concept system. All the predicates are based on real data of a building and work well on SWI-Prolog. The reasoning results are just expected.

Key words: human robot interaction, semantic map, robot navigation, logic programming