Improvements to the Strong Planning Algorithm

摘要: This paper points out some drawbacks and gives modifications to the strong planning algorithm. In particular, a set of states is found by using adjacency matrix that corresponds non-deterministic state-transition system. The composed four parts, first part initial which can not reach goal states, if empty, there planning; state states,we put into second part; third are unreachable from states; so station-action pairs relate or absolutely useless; fourth without passing any before, state-action useless as well, because they move execution away goal. So great many be eliminated directly universal policy. Finally, efficiency modified algorithm illustrated an example experiments.