Kevin Choi
Robotics PhD Candidate
Dynamic Adaptive Robotic Technologies
Georgia Institute of Technology

Committee
Dr. Anirban Mazumdar (Advisor), Woodruff School of Mechanical Engineering, Georgia Tech
Dr. Shreyas Kousik, George W. Woodruff School of Mechanical Engineering, Georgia Tech
Dr. Jonathan Rogers, Daniel Guggenheim School of Aerospace Engineering, Georgia Tech
Dr. Matthew Hale, School of Electrical and Computer Engineering, Georgia Tech
Dr. Kyle Williams, Pathfinder Technologies, Sandia National Laboratories

Abstract
Autonomy in fixed-wing airplanes provide valuable services to modern life and have seen increasing use over the recent years. However, autonomous pilots can be inflexible and lack intuitive intelligence compared to their human counterparts. In particular, agility, adaptability, and intelligence of aerial agents become important in adversarial scenarios. This thesis presents specific methods for improving performance in these metrics, organized across three chapters: 1. improving agility through re-planning in primitive-based planners, 2. improving adaptability through a learned model-free policy with Monte-Carlo Tree Search, and 3. improving intelligence by predicting 6DOF aircraft trajectories through recurrent neural networks. Each chapter will introduce a novel algorithm and demonstrate several interesting applications for the new algorithm.