Probabilistic Planning II
The topic of the seminar will be non-deterministic planning. I plan to cover the most relevant aspects (for planning) of the underlying mathematical models used for casting probabilistic planning problems, the standard and novel algorithms to solve these models, and the current limitations and challenges that lay ahead for developing successful planners capable of tackling large and complex problems. As we will see, the underlying mathematical models are complex and well understood, there are many algorithms for computing solutions, but very few heuristics. Thus, the field offers many opportunities to young and enthusiastic people looking to do groundbreaking research not only in planning but also in areas such as stochastic control theory and operations research.