AUTONOMOUS CONTROL IN BEETLEWEIGHT COMBAT ROBOTICS USING COMPUTER VISION

Abstract

Beetleweight Combat Robotics is an event-driven sport consisting of two or more opposing, three pound robots in an enclosed arena, traditionally remotely controlled by a pilot. A novel alternative to the standard human pilot is an autonomous control algorithm using computer vision via a top down orthographic camera. A closed-loop control algorithm is implemented, in which robots in an arena are detected and subsequent motion correction is calculated and relayed to the autonomous robot. Autonomous control carries the benefit of removing human reaction time while increasing command frequency, which improves the resolution of control over a standard human pilot. A Combat Robotics arena provides a consistent testing ground for autonomous control where com plex object interaction is analyzed with computer vision. Consequently, this experience enables motivated engineers in the Wildcat Robotics Club at the University of Arizona to gain experience with computer vision and develop the skills required to advance the state of self driving technology in motorized vehicles.