Influence Diagrams for Optimal Decision Making in Early Science Space Mission Concept Development

Concept design maturation is an initial phase in theengineering design process that revolves around decisionmaking. In the context of science space mission conceptdevelopment, decision trees play a crucial role. The branches indecision trees are interdependent, meaning that the probabilityof future outcomes depends on past decisions. However, thereare limited resources available for exploring these differenttrade space options, such as time, money, and workforce.Therefore, the main challenge is to efficiently conduct this tradespace exploration and generate a wide range of mission ideasand alternatives from which new programs/projects can beselected. This is the purpose of the NASA Pre-Phase A projectlife cycle phase.To better understand the concept of decision trees, it'simportant to clarify certain terms. A decision tree is a specifictype of graph, which is a collection of objects where some objectshave relationships with each other. In a directed graph, theseobjects are called vertices and are connected by directed edgesor arcs. A directed acyclic graph (DAG) is a directed graph thatcontains no cycles, meaning there are no loops formed whenfollowing the directions of the edges.One approach to reaching optimal decisions is by employing aninfluence diagram. An influence diagram (ID) is a concisegraphical and mathematical representation of a decisionscenario. By utilizing the structure of a directed acyclic graph,influence diagrams provide a visual and intuitive way torepresent the relationships between variables, decisions, andoutcomes, facilitating analysis and decision-making underuncertainty. Influence diagrams are particularly useful formodeling decision problems that involve uncertainty andcomplex dependencies. They can represent decision problemsinvolving multiple decision-makers, allowing for a morecomprehensive analysis of interactions and dependenciesbetween different decision-makers. Influence diagrams can alsohandle decision problems with multiple objectives, enablingdecision-makers to consider trade-offs between differentcriteria and incorporate subjective preferences. Influencediagrams extend the concept of a Bayesian network and allowfor modeling and solving both probabilistic inference problemsand decision-making problems based on the maximum expectedutility criterion.This paper will explore the development of influence diagramsand their application in science space mission conceptdevelopment at the Jet Propulsion Laboratory.