2019-learned-path-collision

Autonomous driving has captured academic and public imaginations for years. This project attempts to implicitly teach a car to follow the best optimized route to a destination while avoiding obstacles. The car is taught the optimized route based on a reward/penalty system via reinforcement learning. Using only the distance away from the nearest object and the angle of said object, the car avoids collisions and learns the optimized route in computer simulated worlds.

This video describes my experience with the CAT Vehicel REU program at the University of Arizona during the 2019 summer. During this program, I realized I enjoy making games for reinforcement learning and what I wanted to research in graduate school. I would recommend this program due to the great mentors who supported me and the friends I made over the summer.