摘要
为提高机器人对家庭环境的语义认知能力,提出了一种环境语义认知模型.基于卷积神经网络提取场景类别和物品语义信息,实现环境的上下文语义理解;采用语义解析器解析语义信息,进而以结构化的方法表征知识,并将其存入本体知识库.同时基于语义网规则语言(SWRL)推理规则挖掘出其隐藏知识,实现知识库的扩展;以服务任务为驱动,实现家庭环境下机器人快速、准确地从知识库获取所需语义信息,从而智能化地执行服务任务.实验结果表明:所提出的模型能够准确提取环境语义信息,并能使机器人快速检索到服务所需知识.
In order to improve the robot’s semantic recognition ability to family environment,a semantic model of the environment was proposed.The model was based on the convolution neural network to extract the scene category and the semantic information of the objects,to realize the context semantic understanding of the environment.Then,in order to make knowledge characterized in a structured way,semantic resolver was used to analyze semantic information,after that it was deposited into the ontology knowledge base.At the same time,based on the SWRL(semantic web rule language) inference rules,the hidden knowledge was mined to expand the knowledge base. Driven by the service task, robots can get the semantic information needed quickly and accurately from knowledge base under family environment,so as to intelligently perform service tasks.Finally,the experimental results show that the proposed model can recognize knowledge accurately,and the robot can search the knowledge needed quickly.
作者
田国会
王晓静
张营
Tian Guohui;Wang Xiaojing;Zhang Ying(School of Control Science and Engineering,Shandong University,Jinan 250061,China)
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2018年第12期18-23,共6页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金资助项目(61773239)
山东省自然科学基金资助项目(ZR2015FM007)
山东省泰山学者工程资助项目
关键词
环境语义认知模型
卷积神经网络
语义提取
语义解析
知识表征
知识推理
environmental semantic cognitive model
convolutional neural network(CNN)
semantic extraction
semantic parse
knowledge representation
knowledge reasoning
