作　　者： ; ; ;

机构地区： 广西师范大学计算机科学与信息工程学院

出　　处： 《计算机研究与发展》 2009年第11期1917-1924,共8页

摘　　要： 高水平的智能机器人要求能够独立地对环境进行感知并进行正确的行动推理.在情境演算行动理论中表示带有感知行动及知识的行动推理需要外部设计者为agent写出背景公理、感知结果及相应的知识变化,这是一种依赖于设计者的行动推理.情境演算行动理论被适当扩充,感知器的表示被添加到行动理论的形式语言中,并把agent新知识的产生建立在感知器的应用结果之上.扩充后的系统能够形式化地表示机器人对环境的感知并把感知结果转换为知识,还能进行独立于设计者的行动推理,同时让感知行动的"黑箱"过程清晰化. Action and action reasoning is a basic part of human activities. People must execute some actions when they want to complete some tasks. Similarly, robot needs to execute some actions when she accomplishes a task. High-level intelligent robot is required to be able to sense the external environment and do correct reasoning about actions independently. It is needed that the external designer writes out background axioms, sensihg results and related knowledge changes for agent when expressing action reasoning with sensing actions and knowledge in situation calculus action theory. This is a kind of action reasoning depending on the designer. The situation calculus action theory is expanded in proper way, the sensors representation is added into the formal language of action theory, and agent＇s new knowledge producing is based on the results of sensor applications in this paper. A platform is provided in which the following issues can be expressed formally： robot is sensing its external environment; the information obtained by robot＇s sensors is converted to robot＇s knowledge automatically; robot does an action reasoning independent of designer. In this way, the ＂black box＂ process of sensing actions will be clear; robot＇s knowledge will be linked to the results of sensors; knowledge-fluent will be regarded as a dynamic knowledge base; and robot can update the knowledge base by executing sensing actions. Furthermore, robot can make action planning and execute actions independent of designer.

关 键 词： 情境演算 行动推理 感知行动 知道逻辑 知识流

领　　域： [自动化与计算机技术] [自动化与计算机技术]