An Autonomous Mobile Robot Architecture for Outdoor Competitions

Outdoor autonomous navigation and obstacle avoidance are common challenges in robotic competitions, and solving them involves many different sets of skills. In this work, we present a modular architecture for controlling a robot in such tasks. It uses unscented Kalman filter and extended Kalman filter to fuse Global Positioning System, inertial measurement unit and odometry data; a model reference adaptive velocity controller, a proportional controller, and a deformable virtual zone algorithm for reactive obstacle avoidance. We also developed a computer vision target tracking algorithm. In the end, we show the importance of having quality sensors to work with.