Vision-Based Cognition of a Humanoid Robot in Standard Platform Robot Soccer

In the Standard Platform League of RoboCup, teams compete using the same hardware platform — the NAO humanoid robot. The NAO has two cameras, but only one can be used at one time. The camera image is the primary input to the robot’s cognition; the features detected by the vision module are used to self-localize and model objects in the world. In this paper, we focus on how the output of the vision module, the vision features, are used by the localization module to determine the robot’s pose, and by the world model module to model objects in the world such as the ball and goal.