Robo-Erectus: A Soccer-Playing Humanoid Robot

Abstract. This paper provides a brief description of a low-cost soccer-playing humanoid robot Robo-Erectus (RE) developed at Singapore Polytechnic. We had participated in the Humanoid League 2002 being held for the first in history, and won second place for the Humanoid Walk competition. To develop a lowcost humanoid platform which is affordable for many researchers, our RE humanoid robots have been intentionally designed to have low-torque servo motors and low-precision mechanical structures so that the cost can be significantly reduced. Inspired by human’s remarkable capability of utilizing perceptions, we conduct a research aimed at synthesizing humanoid gaits through the incorporation of both perception-based and measurement-based information. For this reason, a fuzzy reinforcement learning (FRL) agent has been developed to implement perception-based humanoid walking and kicking pattern generation for the Robo-Erectus.