Optimization-Based Decision Support Software for a Team-In-The-Loop Experiment: Asset Package Selection and Planning

This paper presents two domain-independent optimization-based planning algorithms for efficiently allocating assets that are needed to execute a set of interdependent tasks. The first algorithm is an asset package selection module that combines mixed integer programming and an extended Murty's decision space partitioning algorithm that can be used to provide human planners with a set of alternative asset packages that meet individual task requirements, while maximizing task execution accuracy. This asset package selection module was embedded in a decision aid that supported a mixed-initiative team-in-the-loop planning experiment (MOC-1) conducted at the Naval Postgraduate School in March 2009. The experiment examined the effectiveness and efficiency of two different Maritime Operations Center (MOC) organizational structures for conducting a mission planning activity that required the use of scarce resources. The second algorithm is a planning module that integrates weighted length algorithm, asset package selection module, rollout strategy, and a pairwise exchange method to assist experiment designers to set the parametric conditions for the mission planning activity (e.g., asset types and numbers, task requirements, and asset capabilities) and to assure that the tasks as presented to the human planners would, in fact, be achievable to a specified level of accuracy.