Co-Explore: Adaptive Goal-based Cognitive Architecture for Human

Metron proposes to develop an adaptive and interactive cognitive system to address the technical challenge of providing onboard mission control decision support in deep space exploration missions where crewmembers must make crucial decisions in the absence of terrestrial mission control oversight. However, communications during deep space missions are challenging and impede mission monitoring and interactions with mission systems and personnel. Communication from ground control may take hours, or longer. We propose to build a modular and adaptive goal-based cognitive system with the understanding of spacecraft operations to provide effective mission management and control during periods of no or low communication.Our innovative cognitive architecture fuses goal-driven autonomy with an adaptive computational representation of mission control expert-knowledge to generate and prioritize spacecraft operations, produce goals in response to these operations, and communicate crew tasks and objectives to accomplish the goals. This solution will schedule high-level functions normally performed by human operators, such as logistics management, payload management, maintenance, and personnel health. It employs a flexible, intuitive, and explainable knowledge representation and case-based reasoning that enables the introspection necessary for the crew to query task reasoning. Additionally, our cognitive architecture will employ active learning to adapt and improve over time by incorporating crew feedback and experience into its knowledge.