Reinforcement Learning for Autonomous Vehicles
Abstract
In this project, we implement and deploy reinforcement learning (RL) algorithms for path planning, decision making, and navigation tasks in autonomous vehicles, and compare them to traditional control algorithms. Various control and reinforcement learning algorithms are deployed in a simulation environment to test their performance on various navigation tasks, and will finally be deployed on a robotic car to study the performance of RL in real life autonomous driving tasks. Along the way, the challenges of mapping and localization for the physical car are explored. Furthermore, our results and observations will be used to hopefully establish RL as a viable alternative to control theory for autonomous navigation related tasks.
Citation
Pandey, Amogh (2021). Reinforcement Learning for Autonomous Vehicles. Undergraduate Research Scholars Program. Available electronically from https : / /hdl .handle .net /1969 .1 /194410.