planning and acting in partially observable stochastic domains
Author(s): Leslie Pack Kaelbling, Michael L. Littman, Anthony R. Cassandra
Venue: Artificial Intelligence 101
Year Published: 1998
Keywords: probabilistic models, planning, state estimation
Expert Opinion: This paper provides an easy to understand introduction to MDP and POMDP, which is the basis for understanding Reinforcement Learning and Bayesian Reinforcement Learning ---two learning techniques commonly used to combine planning and learning in robotics. I usually ask beginning research students (honours / 1st year MPhil/PhD) to read this paper.