The 6th edition of the International Probabilistic Planning Competition (IPPC)
will be held in June 2018. We will discuss our submission to the IPPC, which
consists of an extension to the probabilistic planning system PROST.
Instead of the usual heuristic implemented in PROST we determinize the
probabilistic planning task and apply a symbolic planning approach to compute
the best possible reward for each step of the horizon, which will then be used
as the heuristic value to guide the search in the probabilistic task.
We mainly discuss two topics: first, how to deal with the determinization (and
planning) of planning tasks based on RDDL (instead of PPDDL). And second, as the
action costs in the determinized task might be state-dependent, we show how to
apply symbolic planning to such tasks.
Florian Geißer received a Master of Science in computer science from the University of Freiburg in January 2014. Since February 2014 he has been working in the Research Group on the Foundations of Artificial Intelligence headed by Prof. Dr. Bernhard Nebel.