Automated Planners and Planning analysis with Durative Actions

Abstract

Today’s need for efficient and effective automated planning has brought about a distinct set of planning approaches focused on different quantifiable results, namely time and resource consumption. The planning problems are diverse and touch many aspects of day-today life. Whether it is for finding the shortest route of travel in presence of traffic to selection of investment strategies, planning routines are in place to meet the needs of everyone and every machine. Most existing planning domains contain actions, which have common parameters such as duration, timing, resource consumption, and in some cases resource generation. This master thesis seeks to find the most optimal planner to solve Logistics problems. Logistics problems utilize time, space, resources and actions thus they represent an ideal case to assess planners and underlying algorithms in how they deal with complexity of issues, how much time they utilize in generating a plan and how many steps they use in reaching the goal. The combination of these factors points to the optimal planner suitable for logistics planning. Planners are reviewed, analysis is undertaken and the most optimal solution is identified through practical experiments.