A process and tool-set for the development of an interface agent for use in the RoboCup environment.

摘要

A tool-set and methodology was created in order to improve a basic interface that was developed to allow a human to act as a player in the RoboCup soccer environment. These tools allow play recognitions to be learned in a layered approach from previous teams and incorporated into an advanced interface that increases the reaction time of a human player. The end result is an iterative process that can be used to incorporate further learning capabilities into such an interface. With the addition of more layered skills and recognitions, it is believed that a team of humans should be able to compete effectively against a team of computer agents. At this point, the interface can be used to learn from human players instead of their computer counterparts.