一种多Agent的情境装配模型  

Multi-agent based context-aware assembly model

作　　者：宋晨[1] 刘惠义[1] 

机构地区：[1]河海大学计算机与信息学院智能科学与技术研究所,江苏南京211100

出　　处：《计算机集成制造系统》2016年第5期1205-1210,共6页Computer Integrated Manufacturing Systems

摘　　要：为了解决虚拟装配系统普遍存在的人机交互不够友好、配置文件复杂、难以应对需求变化等问题,提出一种以多Agent技术为基础的情境装配模型,将每个虚拟装配零件视作独立的Agent,构建一个自底向上的情境装配场景。首先从装配体的视角构建装配信息模型,刻画零件对象和特征对象的各项参数,同时定义用于Agent之间传输不同状态参数的消息模板,并根据虚拟装配任务的过程特点建立了树形决策机制,装配体根据收到的消息及其中包含的参数决定下一步装配。通过实际项目应用验证了该模型的有效性。To solve the problems such as low efficiency of man-machine interaction and disability to cope with require- ment changes caused by complex configuration and so on in virtual assembly systems, a virtual assembly model based on Agent was proposed, which constructed a bottom-up virtual assembly scene by considering each assembly part as an individual agent. From the viewpoint of each part, the assembly knowledge model was constructed, which included some parameters for describing parts and features. Meanwhile, serials of message templates for transfor- ming context information between agents were also defined. To help agents make proper decision, a decision tree was built based on the process of virtual assembly, by which agent could choose its next step with received messages and relevant parameters. Experiment showed that the proposed assembly model could accurately understand the users＇ intention according to the context-aware.

关 键 词：虚拟现实 虚拟装配 情境装配 多AGENT系统 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]