This talk introduces Auto-tuned Primal-Dual Successive Convexification (Auto-SCvx), an algorithm designed to reliably generate dynamically feasible trajectories for constrained hypersonic reentry problems. The method solves a sequence of convex subproblems that converge to a solution of the original nonconvex optimal control problem. A key feature of Auto-SCvx is its closed-form update of dual variables, which adaptively tunes constraint penalty weights, eliminating the need for manual hyperparameter tuning. As NASA and commercial space programs increasingly pursue human-rated and reusable launch vehicles, reentry trajectory planning has become a critical application. Developing efficient, reliable algorithms for these sensitive problems—where strong coupling often exists between mission parameters and algorithm hyperparameters—is essential to enabling robust, scalable solutions.
Skye Mceowen, Researcher & PhD Candidate | UW Autonomous Controls Lab
Skye Mceowen is a Ph.D. candidate and researcher in Aeronautics and Astronautics at the University of Washington, specializing in convex optimization-based trajectory planning for aerospace vehicles. Her work focuses on sequential convex programming, particularly the Auto-SCvx algorithm—an auto-tuned primal-dual successive convexification method—applied to both aerial quadrotor drones and hypersonic reentry vehicles. This research earned her the 2025 AIAA SciTech Graduate Student Paper Award. Skye previously managed the Autonomous Controls Lab’s physical research space and has had multiple internships across the aerospace industry over the past decade, including roles at SpaceX and Blue Origin.