基于本体的物品属性类人认知及推理

Ontology-based humanoid cognition and reasoning of object attributes

为了实现服务机器人对物品知识的类人认知和推理,以辅助机器人执行物品操作任务,提出基于属性的家庭物品语义知识库构建方法.通过对物品知识进行分类,使用视觉、类别、物理、功能、操作属性对物品知识进行系统化和条理化的描述,结合本体建模语言对物品知识进行建模;利用语义网络规则语言,制定针对物品属性的推理规则(包含视觉、类别、物理—功能、视觉—类别、类别—类别),实现对物品属性的自主认知及推理.在实验室搭建的机器人物品操作仿真平台上验证了物品语义知识库对机器人操作物品的指导作用.对语义知识库的查询结果表明,机器人可以以统一的形式获取与物品相关的多种高层语义信息.

A method to construct a semantic knowledge base based on object attributes was proposed in order to help robots have humanoid cognition and reasoning about object knowledge to assist them to operate objects. Visual attributes,category attributes,physical attributes,affordance,operation attributes were used to describe object knowledge in a systematic and orderly way,and web ontology language was used to model the object knowledge.The rules between attributes were made by semantic web rule language to fulfil the cognition and reasoning about object attributes.A simulation platform was used to verify that the knowledge base can guide the operation of objects.The query results to the knowledge base show that robots can obtain different kinds of semantic information in a unified form.