Machine Learning Applications for Spacecraft Mission Design in Complex Environments

The field of spacecraft dynamics and trajectory design has not been thoroughly explored through the lens of machine learning. While a number of other industries and even other aspects of the aerospace field have benefitted greatly from the expanding field of machine learning, spacecraft mission design has yet to fully reap these advancements. Thus, this project will investigate both modeling the dynamics of a spacecraft as well as control the spacecraft’s trajectory through machine learning techniques. First, a stable of machine learning techniques and a data processing architecture will be built up to quickly run through experiments on various astrodynamics problems. Concurrently, machine learning algorithms will be trained to extract the dynamics and develop some understanding of restricted N-body dynamics, especially looking at techniques and setups commonly used by mission designers. After learning these dynamics, a control method, most likely some form of reinforcement learning, will be developed to create an autonomous mission designer. By first learning the dynamics, it is expected that stronger algorithms for designing and predicting the reward function will be able to be developed as a result. Additionally, an expected by-product of this work will be a quicker method of predicting chaotic dynamical systems when compared against typical integration techniques, at least to some finite horizon. Approaching this problem from a machine learning perspective will allow for a quicker, robust manner to generate candidate trajectories and more fully explore the design space for mission concepts.

This is early/mid-stage (TRL 3) — the most realistic path in is NASA SBIR/STTR, which funds small businesses and research institutions to develop technology aligned with NASA's needs (equity-free, phased funding). Check whether a current SBIR/STTR solicitation topic overlaps with this project's technology area, or contact the project directly (above) to ask.

None of these are guaranteed paths for this specific project — TechPort itself doesn't have an "apply" button. Reaching out to the contact(s) above with a specific question is usually the fastest way to find out what's actually open.