Optimal Control for Space Systems

3 Credits.

In this course, students will learn how optimal control can used to model and solve model-based reinforcement learning problems arising in aerospace and robotic applications. In particular, a strong emphasis is placed on real-time planning and control via the use of on-board numerical optimization and students will apply theoretical insights from trajectory optimization and model predictive control for developing custom solvers leveraging modern automatic differentiation libraries. Specifically, students will apply theory to practice by implementing custom interior point method solvers, sequential quadratic programming, and differentiable optimization techniques. Applications studied will include planetary rover path planning, rocket powered descent guidance, and spacecraft controls. Finally, a course project will be included to allow students to gain further experience on an algorithm or application of their choice.