Racing to build your AI skills: programming RL models with AWS DeepRacer

Abstract

Amazon DeepRacer provides an innovative way to integrate complex AI/ML knowledge into a hands-on learning experience for students. Competitions provide a challenging pedagogical way to foster collaborative spirits in students. This study assesses the impact on the knowledge and confidence of students participating in groups in a DeepRacer competition at a small private school in Western Pennsylvania. Students were enrolled in teams and worked with faculty advisors to train their models and for a DeepRacer competition. A pre-posttest assessed their knowledge and confidence level in the AI/ML model training. The results suggest a significant improvement in the knowledge and confidence of the students about reinforcement learning. When used as a pedagogical tool, these competitions motivate students to apply theoretical knowledge to real-world situations and enhance their problem-solving abilities. Implications are drawn, and limitations are noted.