Hierarchical Task Planning under Uncertainty

In this paper we present an algorithm for planning in nondeterministic domains. Our algorithm C-SHOP extends the successful classical HTN planner SHOP, by introducing new mechanisms to handle situations where there is incomplete and uncertain information about the state of the environment. Being an HTN planner, C-SHOP supports coding domain-dependent knowledge in a powerful way that describes how to solve the planning problem. To handle uncertainty, belief states are used to represent incomplete information about the state of the world, and actions are allowed to have stochastic outcomes. This allows our algorithm to solve problems involving partial observability through feedback at execution time. We outline the main characteristics of the algorithm, and present performance results on some problems found in literature.