Reinforcement Metalearning for Interception of Maneuvering Exoatmospheric Targets with Parasitic Attitude Loop

What is it about?

We use reinforcement meta-learning, i.e. a machine learning technique for sequential decision making, to optimize and integrated GNC system for exoatmospheric target intercept. The proposed system is designed to "learn-to-learn", i.e. it is trained on a distribution of different environments so that it can quickly adapt to situations never seen before.

Featured Image