Semester of Graduation
Electrical and Computer Engineering
First Major Professor
Master of Science (MS)
Navigation and object interaction are two difficult, crucial tasks for autonomous mobile vehicles. These tasks are made even more challenging when the vehicle is in an unfamiliar environment. The task of autonomous refueling in an arbitrary environment encompasses both the navigation and object interaction tasks. In this paper, we propose a reinforcement learning model and training procedure which can take the first steps toward efficiently learning to seek out and dock at a charging station using only a single on-board monocular camera. More specifically we address the task of rotating in place to find the charging station and keep it centered in the camera's field of view. Our results show that using this method, the vehicle is able to successfully find and maintain focus on the charging station with a success rate of 99.4%.
Gilbert, Nathaniel, "Using Reinforcement Learning to Tackle the Autonomous Recharging Problem" (2019). Creative Components. 389.