TITAN: A Novel Trust and EffIcienT AI-based Hierarchical HeterogeNeous Planning and Scheduling at Tactical Edge

Military decision making takes place in a variety of complex domains (defense, security, cyber, etc.). Artificial Intelligence not only allows for data reduction and synthesis, but also for the development of predictions about future events, and about outcomes to considered interventions. However, due to uncertain circumstances and ill-defined problems, AI cannot yet do this autonomously. In this proposed effort, IFT and its academic partner, the University of Nevada, Reno, seek to develop technical underpinnings and methodologies for enhancing and modernizing autonomous/automated decision making in multi-domain. We will first develop a novel hierarchical heterogeneous Markov model that can provide a solid theoretical foundation to build heterogeneity of various domains and a hierarchical decision-making pipeline within each domain. Then, we will develop a novel hybrid actor-critic reinforcement learning (ACRL) technique that can learn the optimal multi-domain planning and scheduling policies within a feasible time. Next, a novel trust AI mechanism will be developed along with Quality-of-Performance (QoP) metrics to balance the optimal efficiency, computation efficiency, and operational risk. In addition, standard simulation tools, e.g. OpenAI Gym and PySC2, and the lab-developed reconfigurable testbed will be used to validate the proposed intelligent hierarchical heterogeneous planning and scheduling technique.