Building high-fidelity gravity models for spaceflight, planetary science, and geodesy using physics-informed neural networks.
How can we learn the dynamics of novel environments with limited data?
How can we reduce the burden on satellite operators to detect and diagnose faults?
Use sparse images taken from safe surveillance orbits, construct a 3D shape model of a target spacecraft or debris.
How can we construct more efficient coordinates and basis functions to represent complex dynamical systems?
Explore how to safely navigate exotic environments in deep space with autonomous decision making agents with reinforcement learning.