A NEW GRAPHPLAN METHOD TO IMPLEMENT PARTIAL GOALS
Several new concepts are introduced, such as single satisfaction, synthesis satisfaction and sub-optimal plan. With those concepts a new algorithm based on Graphplan is presented. A lot of restrictions make the Graphplan planner far from the real world. The new algorithm inherits all the capabilities of Graphplan. Except this, it also adapted to other situations. For a problem, if Graphplan cant find a solution implementing all the goals, the new method will output a best sub-optimal solution, which can implement part of goals. And these partial goals can satisfy user as much as possible. In many complex circumstances, relaxing partial rigid restrictions can solve a problem more expressive and more effective.
Partial goals Graphplan Single satisfaction Synthesis satisfaction Sub-optimal plan
JIA-YUE REN WEI-ZHOU GUAN FANG WANG GANG HOU
School of Computer, Northeast Normal University, Changchun 130117, Jilin, China
国际会议
2006 International Conference on Machine Learning and Cybernetics(IEEE第五届机器学习与控制论坛)
大连
英文
814-818
2006-08-13（万方平台首次上网日期，不代表论文的发表时间）