Published Nov 24, 2023

734: Humanoid Robot Soccer — with the Dutch RoboCup Team

Dive into the world of humanoid robot soccer with Jon Krohn and Dario Catarrinho as they reveal how the Dutch RoboCup Team utilizes cutting-edge machine learning techniques, including vision and reinforcement learning, to create autonomous robots that excel in global competitions.
Episode Highlights
Super Data Science: ML & AI Podcast with Jon Krohn logo

Popular Clips

Episode Highlights

  • Physical Design

    The humanoid robots used in the RoboCup soccer matches are about two feet tall and bipedal, resembling human players in miniature form. These robots, known as Nao, are fully autonomous and equipped with cameras for ball detection, utilizing convolutional neural networks to identify the ball on the field 1. The matches take place on a standard-sized indoor turf, approximately 15 meters long, where teams from around the world compete annually. explains, "These robots do this fully autonomous, so there's no human intervention."

       

    Game Dynamics

    The dynamics of robot soccer involve complex algorithms that allow the robots to respond to game events like whistles and goals. describes how the robots use sound wave detection to differentiate a whistle from ambient noise, enabling them to reposition after a goal is scored 2. The robots play in teams, with matches ranging from five to seven robots per side, showcasing their ability to autonomously navigate and compete. "These robots are pretty clumsy because they really aren't made to play football," Dario notes, highlighting the challenge and excitement of the game 2.

Related Episodes