Proceedings Abstracts of the Twenty-Fifth International Joint Conference on Artificial Intelligence

Probabilistic Planning with Risk-Sensitive Criterion / 3996
Ping Hou

While probabilistic planning have been extensively studied by artificial intelligence communities for planning under uncertainty, the objective to minimize the expected cumulative cost is inappropriate for high-stake planning problems. With this motivation in mind, we revisit the Risk-Sensitive criterion (RS-criterion), where the objective is to find a policy that maximizes the probability that the cumulative cost is within some user-defined cost threshold. By combining goal-directed MDPs and POMDPs with the RS-criterion, the corresponding risk-sensitive probabilistic planning models —Risk-Sensitive MDPs (RS-MDPs) and Risk-Sensitive POMDPs (RS-POMDPs) — can be formalized. The overall scope of this research is to develop efficient and scalable RS-MDP and RS-POMDP algorithms.