Automatic Generation of Efficient Domain-Optimized Planners from Generic Parametrized Planners. In: 18th RCRA workshop on Experimental Evaluation of Algorithms for Solving

When designing state-of-the-art, domain-independent planning systems, many decisions have to be made with respect to the domain analysis or compilation performed during preprocessing, the heuristic functions used during search, and other features of the search algorithm. These design decisions can have a large impact on the performance of the resulting planner. By providing many alternatives for these choices and exposing them as parameters, planning systems can in principle be configured to work well on different domains. However, usually planners are used in default configurations that have been chosen because of their good average performance over a set of benchmark domains, with limited experimentation of the potentially huge range of possible configurations. In this work, we propose a general framework for automatically configuring a parameterized planner, showing that substantial performance gains can be achieved. We apply the framework to the well-known LPG planner, which has 62 parameters and over 6.5 × 10 possible configurations. We demonstrate that by using this highly parameterized planning system in combination with the off-the-shelf, state-of-the-art automatic algorithm configuration procedure ParamILS, the planner can be specialized obtaining significantly improved performance.