With the expansion of autonomous robotics and its applications (e.g. medical, competition, military), the biggest hurdle in developing mobile robots lies in endowing them with the ability to interact with the environment and to make correct decisions so that their tasks can be executed successfully. However, as the complexity of robotic systems grows, the need to organize and modularize software for their correct functioning also becomes a challenge, making the development of software for controlling robots a complex and intricate task. In the robotics domain, there is a lack of reference software architectures and, although most robot architectures available in the literature facilitate the creation process with their modularity, existing solutions do not provide development guidance on reusing existing modules. Based on the well-
known IBM Autonomic Computing reference architecture (known as MAPE-K), this work defines a refined architecture following the Robotics perspective. To explore the capabilities of the proposed refinement, we implemented the RoCS (Robotics and Cognitive Systems) framework for autonomous robots. We successfully tested the framework under simulated robotics scenarios that mimic typical robotics tasks and evidence the framework reuse capability. Finally, we understand the proposed framework needs further experimental evaluation, particularly, assessments on real-world scenarios.