Self-localization of humanoid robots with fish-eye lens in a soccer field

Abstract

The robot localization problem has been studied for decades and the particle filter algorithm has been successfully applied for the localization of wheel-based mobile robots. In this paper, we provide an implementation of the particle filter algorithm for the self-localization of our soccer playing humanoid robots, RO-PE (RObot for Personal Entertainment) VI, which participate in the Humanoid League (kid-size) of RoboCup in the past few years. The localization algorithm enables the robot to localize itself with respect to the soccer field. For our robot platform, there are a few challenges, such as the locomotion system has high variability in achieving the actual displacement; the vision system employs a fish-eye lens which has significant distortion and suffers from the oscillation caused by the locomotion; and limited computational power. In this paper, we propose a series of solutions for all these challenges and show their effectiveness by implementing the algorithm on the physical robots.