不确定规划中的观察信息约简方法及其应用研究

项目名称： 不确定规划中的观察信息约简方法及其应用研究

项目编号： No.61272295

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 自动化技术、计算机技术

项目作者： 文中华

作者单位： 湘潭大学

项目金额： 80万元

中文摘要： 在不确定规划中进行观察信息约简有重要的理论意义和广阔的应用前景，可以降低规划执行的成本和提高规划执行的效率。本项目首先找出求弱规划和强循环规划解的问题所需要区分的状态对集合，然后设计在全可观察条件下求弱规划和强循环规划解的观察信息约简方法。对于给定的一个部分可观察的弱规划和强循环规划问题，首先假设所有的状态变量都不是观察变量，通过逐步增加观察变量来设计在部分可观察条件下求弱规划和强循环规划解的观察信息约简方法。现实中获得不同观察信息所需要的代价往往是不同的，通过构造整数规划约束不等式组设计找一个代价最小观察组合方案。为避免传感器故障造成的严重后果，保证在部分传感器出错时仍可以判定当前系统的状态，以提高整个系统的可靠性，设计找出容错最优观察集的方法。利用上面所述成果，设计在机器人路径规划和多机器人协同规划中进行观察信息约简的方法，以便机器人作出快速反应，协同工作，抢得先机，提高效率。

中文关键词： 观察信息约简；协同规划；权值；多机器人；机器人路径规划

英文摘要： Obseration reduction has important theoretical significance and potential applications in a non-deterministic planning, it can reduce costs and improve efficiency of planning. The project first found state pairs which are requested distinguish in the weak planning and strong cyclic planning solution. And then a method has been designed to reduce observation of the weak planning and strong cyclic planning. Given a partially observable weak planning and strong cyclic planning problem, first assume that all state variables are not observed variables, a way which increases the observation variables gradually is used for the observation reduction in the partially observable conditions weak planning and strong cyclic planning solution. The reality observation information need to cost which are different, we design a method that constructs an inter constraint inequalities to find a minimal cost observe. A method is designed To avoid the serious consequences of sensor failure in order to improve the reliability of the entire system. A method has been designed to reduce observation in the robot path planning and the multi-robot collaborative planning in order to respond quickly to work together and improve efficiency.

英文关键词： Observation Information Reduction；Collaborative Planning；Weight Value；Multiple Robots；Robot Path Planning