Towards Incorporating AI into the Mission Planning Process

Abstract

AbstractWhile there are numerous powerful tools to support Navy mission planning, the mission planning process still remains a hybrid planning activity across human operators and advanced tools. Advances in artificial intelligence (AI) have seen an increase in interest and use in the mission planning environment. Yet traditional approaches typically focus on optimizing the performance of the individual operator or the mission planning tool, not the joint problem solving that is needed for the human and AI team necessary in these envisioned mission planning environments. For example, while AI-based approaches can offload extensive processing by the human operator, outcomes of AI-based tools are seldom presented in ways that make it readily understand by a human or fit in with the overall process. This results in new and additional work for the operator as they must manually translate these outcomes into actionable information. There are also specific characteristics of AI-based approaches to consider, such as accounting for data limitations; many mission planning environments have data availability constraints or do not capture the right data. New methods are needed to specifically consider the benefits of AI-based approaches, ensure that AI derived insights are communicated effectively to operators, and optimally support operators in their own mission planning workflow.In this paper, we describe an overall approach for incorporating AI into the mission planning process. First, to consider how best to leverage the unique capabilities and constraints of a human operator and artificial intelligence technologies as a cooperative team throughout the mission planning process, we will frame the human operator and AI system as a single unit and take a joint cognitive system (JCS) analysis approach. This approach will identify the key information to jointly create the most effective mission plan most efficiently. Next, we describe a novel approach, neural policy programs (NPPs) to address some of the critical challenges of incorporating AI into the mission planning process. Finally, we describe our approach for extending to the StarCraft II game and creating a pre-game planning stage to prototype our approaches.KeywordsMission planningArtificial intelligenceJoint cognitive systems (JCS)Neural policy programsStarcraft II