Abstract: This paper presents an extension to a partially observable Markov decision process so that its solution can take into account, at the beginning of the planning, the possible availability of free information in future time ...