Adapting Computer Vision Algorithms to Smartphone-based Robot for Education

Abstract

Robobo is a smartphone-based educational robot used on multiple education stages. The following paper describes the development and testing of three computer vision algorithms that run natively in Robobo. They perform real-time object identification, ArUcO marker detection, and lane detection. With them, the students that use the Robobo robot can develop highly realistic projects on mobile robotics. This brings robotics teaching closer to real robotics, where advanced computer vision algorithms are widely used. The characterization of the algorithms that are presented in this document shows that their performance is very high in comparison to implementations that run on a standard computer.