An important aim of the current research e ort in artificial intelligence and robotics is to achieve cooperative agent behavior for teams of robots in real-world scenarios. Especially in the RoboCup scenario, but also in other projects like Nexus, agents have to cooperate eficiently to reach certain goals. In the RoboCup project, cooperative team-play and team strategies similar to real world soccer are intended. This article describes an approach that combines cooperative aspects, role assignment algorithms and the implementation of robot behavior with Interaction Nets. Based on these single methods, a complete framework for team strategies was developed that is used in the RoboCup environment for the middle-size team CoPs Stuttgart and in the Nexus project, where a team of robots guides persons through a building.