MOMDPs: A Solution for Modelling Adaptive Management Problems

In conservation biology and natural resource management, adaptive management is an iterative process of improving by reducing uncertainty via monitoring. Adaptive the principal tool for conserving endangered species under global change, yet problems suffer from a poor suite solution methods. The common approach used to solve problem assume system state known dynamics can be one set pre-defined models. method unsatisfactory, employing value iteration on discretized belief MDP which restricts study very small problems. We show how overcome this limitation modelling as restricted Mixed Observability called hidden model (hmMDP). demonstrate simplify function, backup operator update computation. that, although simplified case POMDPs, hm-MDPs are PSPACE-complete in finite-horizon case. illustrate use manage population threatened Gouldian finch, bird endemic Northern Australia. Our simple important step towards efficient algorithms solving