Opportunity and Challenges: 

• Reinforcement Learning (RL) enables training of an artificial intelligence (AI) agent to operate in dynamic uncertain environments 

• Impressive performance outcomes to learn nearly-optimal solutions in a variety of application domains 

• Particularly important for high-speed aerospace missions where real-time trajectory generation is computationally prohibitive 

• Limited testing and characterization of performance bounds of RL solutions 

• Impedes transition to real time systems 

Technical Contributions:

• Develop a comprehensive Test and Evaluation Framework for Reinforcement Learning  

• • Robustness Testing of RL solutions  

• • Understanding of RL decision making via Explainable AI 

• • Validation of RL solutions  

• Demonstrate application of RL to a high-speed aerospace vehicle mission 

• • Investigate uncertainty in flight parameters such as angle of attack, velocity, altitude, and flight path angle