One of the most important issues in the physical agent challenge is the design of the standard platform and their supply because the range of the physical agent designs is to too wide to evaluate the skills achieved in the same level. If we had a precisely specified platform for RoboCup, the situation would be similar to the current simulation track because the design issues of mechanical structure would disappear. On the other hand, completely free design makes it hard to evaluate the skills achieved on equal terms. As a middle ground, we would like to have a platform that allows us to easily reconfigure the physical structure by which we expect physical agents to develop complex behaviors.
Recently, an open architecture called OPENR has been proposed as such a platform [Fujita and Kageyama1997]. OPENR has 1) standard interfaces between physical and software components and a programming framework, 2) configurable physical components with a common interface and information exchangers of their function and configurations, and 3) is constructed as a layered architecture based on object-oriented robot OS.
Expecting that an open architecture such as OPENR will be available for the RoboCup context in the very near future, we will offer basic resources and opportunities in order to facilitate technical progress based on the RoboCup physical agent challenge.