Recent advances in planning techniques for partially observable Markov decision processes (POMDPs) have focused on online search techniques and offline point-based value iteration. While these techniques allow practitioners to obtain policies for fai...