Our project titled “Explainable AI for Mission Planning and Execution with Interpretable Courses of Action” is selected for funding by the ONR Science of AI program! Navy mission planning and execution requires the ability to consider a large number of potential courses of action (COAs) and ensure optimal decision-making in uncertain environments. We aim to address these challenges by: (1) developing methods to classify decision sequences from AI systems into clusters of coherent and interpretable COAs; (2) developing methods to discover and incorporate commander biases into AI decision policies; and (3) developing a methodology to evaluate and optimize hierarchical reinforcement learning (RL) architectures with respect to explainability, adaptation, and mission performance.

This project is a collaboration with Dr. Girish Chowdhary, Dr. RS Sreenivas, Dr. Jaggi Yedetore, and Dr. Heather Filippini from UIUC.