|
In recent years, Intelligence Planning,especially uncertainty planning problems has became a very hot focus in Artificial Intelligence.Some researchers believe that possibilistic theory is more suitable for cases in which the probabilities are not available,not reliable,or hard to obtain,because of the ability of describing imprecise and vague information under uncertain environmen, possibilistic planning is more suitable to the real world.But existing possibilistic planner performs under the restriction of allowing only one non-noop action per time step, mutex can not work in the planner, the solution is very long and cost much time and space.On the basis of in-depth study of the theory of probability,to solve this problem, we proposed a new algorithm PC-Graphplan in this paper.Firstly, we give a new concept of mutex under possibilistic graphplan, to enable it to meet the uncertainty of the actions in the possibilistic planning, PC-Graphplan sign and propagate the information of mutex with planning graph, and improve the quality of solution.Secondly,We give the concept of concurrent actions,and find a optimistic algorithm and a pessimistic algorithm to generate the action set, realize executiving concurrent actions per time step,breaches the restriction“allows only one non-noop action per time step”in poss-Grahplan.Thirdly,we give a new algorithm called PC-Graphplan based on optimistic and pessimistic points which can handle the problem of concurrent actions in the frame of possibilisitic graphplan. At the same time,we have developed a new planner--PC-graphplan,which realizes handling concurrent possibilisitic planning,we compare it with Poss-Graphplan in to domains,and find it more suitable for actual problems .
|