传感器机器人行动推理的有色网表示模型  

A Representation Model for Reasoning about Actions of Robot with Sensor

作　　者：钟珊[1] 常晋义[1] 乐德广[1] 何世明[1] 

机构地区：[1]常熟理工学院计算机科学与工程学院,江苏常熟215500

出　　处：《计算机仿真》2011年第3期242-245,共4页Computer Simulation

基　　金：江苏省"六大人才高峰"高层次人才项目(09-D-LE-Deguang)

摘　　要：研究机器人行动推理优化系统,针对传统的推理前必须预先给定所有环境状态,不能动态获取环境状态新知识。为了使得机器人在推理的过程中能动态获取环境状态新知识以提高推理的准确度,提出了行动推理过程中的两种基本动作即外部动作和感知动作进行了形式化地表示,对这两种基本动作、STRIPS推理规则以和有色网来表示机器人在不完全可知环境下进行行动推理的形式化表示,采用PNS(Petri Net for Reasoning about Action with sensor)网系统,采用CPNTools对办公环境下机器人行动推理实验,结果表明PNS网系统能使得机器人在行动推理过程中动态获取新知识目标,提高了行动推理的准确度。The representing method for reasoning about action is researched.Gor reasoning needs,the traditional method has to predefine all the environments state,which can not dynamically obtain the environmental new knowledge.In order to improve the reasoning accuracy and make the robot able to get the new state knowledge dynamically,external and internal actions are formalized,and a formalized method of Petri Net for Reasoning about Action with sensor（PNS） is introduced in incomplete environment.The PNS net system of robot office is modeled and simulated by using the CPNTools.The experiment result shows that PNS net system can make robot dynamically obtain the door state and generate the action sequence,and improves the accuracy of reasoning.

关 键 词：有色网 机器人 传感器 行动推理 动作序列 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]