Hands-On Learning in Autonomous Vehicle Development: Outcomes of an NSF REU Program Using Real Vehicles

This paper presents outcomes from a three-year NSF-funded Research Experiences for Undergraduates program focused on autonomous vehicle software development. The program bridged the "sim-to-real" gap in automation education through hands-on projects integrating computer vision, machine learning, autonomous navigation, vehicle-to-everything communication, and smart infrastructure, employing the use of two full-size, street-legal vehicles. Hosted at Lawrence Technological University and run jointly with Michigan State University, the program emphasized skill development, self-efficacy, and research readiness through real-world testing. Our pre- and post-assessment results show that using real vehicles effectively bridged the gap between theoretical knowledge and real-world application. This “reality gap” challenged students to adapt algorithms to dynamic conditions, enhancing their critical thinking and problem-solving skills. External educational evaluations revealed significant improvements in technical proficiency and participant confidence while identifying challenges in diversity recruitment and sustaining graduate interest. Insights gained will inform future STEM initiatives to foster an inclusive, theoretically- and practically-skilled workforce in autonomous systems.