During spacecraft development, information about the system is constantly gained as design and analysis become more detailed. In particular, two types of epistemic uncertainty, design uncertainty and model inadequacy, are always high at the beginning of a project but continuously decrease as the system is designed. However, existing design processes do not typically revisit old design decisions in the context of new information. A risk that can arise from this gap is that the chosen design solution is ill suited for its mission, increasing cost, schedule, and mission failure rate. A strategy to mitigate this risk is to propagate the consequences of new information throughout the current knowledge base, including revising decisions made in an obsolete context. This talk will propose a methodology for system model update using algorithms from incremental planning that address the need for incorporating new information. The Lifelong Planning A* (LPA*) algorithm optimally updates a knowledge base when new information is learned and efficiently finds a new design solution. My proposed methodology adapts the LPA* algorithm for the spacecraft design problem in order to mitigate the risk of design uncertainty or model inadequacy leading to an inefficient spacecraft design and to guarantee that the design solution is consistent with the latest information about the design space.

Biography:

Mark Chodas is a PhD candidate at the Massachusetts Institute of Technology. He graduated from MIT with an S.B. in Aerospace Engineering in 2012 and an S.M. in Aerospace Engineering in September 2014. His research examines model-based systems engineering and the challenges of developing spacecraft in an evolving design space. Mark is also the instrument systems engineer for the REgolith X-ray Imaging Spectrometer (REXIS) instrument on the OSIRIS-REx mission.