(In-person attendance might have better interactive experiences.)

Safe Active Fault Estimation for Spacecraft

 In time-critical operations, rapid and autonomous fault estimation may be necessary to ensure mission safety. Active diagnosis using sequences of planned test actions can resolve ambiguity and provide fast identification of underlying faults. However, naïve test actions could endanger the system. To resolve this, we propose an online tree-based planning algorithm which simultaneously selects actions that gather information about active failures and ensures collision avoidance by satisfying probabilistic chance constraints. We demonstrate the real-time performance of our algorithm on a robotic spacecraft simulator, and show with enough planning, we can reliably maintain safety and diagnose faults.

Tentative Agenda: (All Time PST (GMT -0800)) (US and Canada)

10:30 AM: Check-in, Networking

11:00 AM: Welcome/Introduction, Briefing by Exhibitor

11:10 AM: Presentation and Q&A

12:40 PM: Additional Q&A (if needed) / adjourn / Networking.

James Ragan is a Ph.D. Candidate in the space engineering at Caltech, advised by Prof. Soon-Jo Chung, applying planning to spacecraft in domains where fast autonomous responses are necessary, such as fault estimation in deep-space proximity operations and detecting cyberattacks in formation flying. He has received bachelor’s degrees in aerospace engineering as well as astronomy and physics from the University of Washington.