Completable planning: A curative learning approach to the imperfect domain theory

The complexity of the real world makes a perfect characteriza-tion impossible. Purely deliberative approaches such as classi-cal planning are thus susceptible to unexpected failures. Pre-ventive learning approaches address the imperfect theory problem through the diagnosis of such failures and the deter-mination of fixes to avoid similar failures in the future. Cura-tive learning approaches such as completable planning instead treat fail ures as alternative outcomes and learn alternative plans to the recover from the failures. Through learning, a completable planner learns to plan only for the more likely out-comes in the particular problem distribution it faces. It thus significantly reduces the COS! of disjuncti ve planning.. As a cu-rative learning approach, it IS also better suited to domains where outrlghl fa ilures are unacceptable or failure diagnosis is expensive. making preventive Jearning infeasible. Complet-able planning is a general incremental learning approach whose Sllccess in real world domains may be increased through the integration ot other learning techniques and the. considerat ion of different planning perspectives