Cooperative mapping of an environment by a team of multiple robots is an important problem to advance autonomous robot tasks for example in the field of service robotics or emergency assistance. A precise, global overview of the area the robots are working in, and the ability to navigate this area while avoiding obstacles and collisions between robots is a fundamental requirement for a large number of higher level robot-tasks in those domains. A cooperative mapping, navigation and communication framework supposing unknown initial relative robot positions is developed in this project based on the ROS libraries. It realizes robot displacement, localization and mapping under realistic real-world conditions. Such, the framework provides the underlying functions needed to realize a task of human activity observation in the future. Initially , local maps are individually constructed by the robots using the common gmapping SLAM algorithm from the ROS libraries. The robots are evolving on circles around the scene keeping a constant distance towards it or they can change radius, for example to circumvent obstacles. Local maps are continuously tried to align to compute a joint, global representation of the environment. The hypothesis of a common center point shared between the robots greatly facilitates this task, as the translation between local maps is inherently known and only the rotation has to be found. The map-merging is realized by adapting several methods known in literature to our specific topology. The developed framework is verified and evaluated in real-world scenarios using a team of three robots. Commonly available low-cost robot hardware is utilized. Good performances are reached in multiple scenarios, allowing the robots to construct a global overview by merging their limited local views of the scene.