Development and Implementation of a Hands-On Robotic Course on Computer Vision for Autonomous Vehicles

This paper introduces the development of a robotic course, Computer Vision for Autonomous Vehicle. Derived from VNAV course offered at MIT, this course is designed for senior undergraduate students and graduate students at the University of Oklahoma, where there are a few courses focused on robotics. Our course offers hands-on experience with robotic cars which have been built at low cost, allowing the students to implement fundamental materials in computer vision in a physical minicity environment built at the university. We integrated a popular visual Simultaneously Localization and Mapping algorithm, called Kimera, which allowed students to gain extensive experience with computer vision and visual SLAM in a low-cost 1/10<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">th</sup> scale minicity, mimicking real-world features. We have provided the Docker version of VNAV course which makes the course broadly accessible to students and educators from different backgrounds in computer science and robotics. The final project is designed for students with different levels of experience in robotics and programming, which can be considered for K12 students in the future to introduce them to robotics/computer vision. While there are few teams and research labs at the University of Oklahoma focusing on robotics, this course bridges the gap by introducing a set of assignments in which the students get involved with data collection in a real-world environment and practicing different concepts of computer vision in a physical robotic platform. All the course materials, including the docker image, are open source, which can be utilized by the instructors and students around the world. We have also released the building material and instructions available at: https://github.com/airoulab/CVAutonomousVehicles.git