The RoboCup Physical Agent Challenge: Goals and Protocols for Phase I Minoru Asada 1, Peter Stone s, Hiroaki Kitano 3, Alexis D rogoul 4, Dominique Duhaut 5, Manuela Veloso 2, Hajime Asama 6, and Sho'ji Suzuki 1 1 Department of Adaptive Machine Systems, Osaka University, Suita, Osaka, 565 Japan 2 School of Computer Science, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, Pennsylvania 15213 USA 3 Sony Computer Science Laboratory, 3-14-13 tIigashi-Gotanda, Shinagawa, Tokyo, 141 Japan 4 LIP6, Universite Paris 6, Case 169, 4 Place Jussieu 75252 Paris CEDEX 05, France 5 Laboratoire de Robotique de Paris, U.V.S.Q. 10-12, avenue de l'Europe 78140 Vdlizy, France 6 Institute for Chemical and Physical Research (Riken), 2-1 Hirosawa, Wako, Saltama, 351-01 Japan Abstract. Traditional AI research has not given due attention to the important role that physical bodies play for agents as their interactions produce complex emergent behaviors to achieve goals in the dynamic real world. The RoboCup Physical Agent Challenge provides a good test-bed for studying how physical bodies play a significant role in realizing in- telligent behaviors using the RoboCup framework [Kitano, et al., 95]. In order for the robots to play a soccer game reasonably well, a wide range of technologies needs to be integrated and a number of technical breakthroughs must be made. In this paper, we present three challenging tasks as the RoboCup Physical Agent Challenge Phase h (1) moving the ball to the specified area (shooting, passing, and dribbling) with no, sta- tionary, or moving obstacles, (2) catching the ball fl'om an opponent or a teammate (receiving, goal-keeping, and intercepting), and (3) passing the ball between two players. The first two are concerned with single agent skills while the third one is related to a simple cooperative be- havior. Motivation for these challenges and evaluation methodology are given. 1 Introduction The ultimate goal in AI, and probably in robotics, is to build intelligent systems capable of displaying complex behaviors to accomplish the given tasks through interactions with a dynamically changing physical world. Traditional AI research has been mainly pursuing the methodology of symbol manipulations to be used in knowledge acquisition and representation and reasoning about it with lit- tle attention to intelligent behavior in dynamic real worlds [Brooks, 1991]. On the other hand, in robotics much more emphasis has been put on the issues