Robotic ecologies are networks of heterogeneous robotic devices prevalently embedded in everyday environments, where they cooperate to perform complex tasks. Although their potential makes them increasingly popular, they all share a fundamental problem. The problem is how to make them both autonomous and adaptive in order to reduce the amount of preparation, pre-programming and human supervision required in real world applications. For this purpose, RUBICON is a special project that develops learning solutions, which yield cheaper, adaptive and efficient coordination of robotic ecologies. In the study, we pursue an approach that builds upon a unique combination of methods from cognitive robotics, machine learning, planning and agent-based control, and wireless sensor networks. This paper elucidates the innovations advanced by RUBICON in each of these fronts, then goes on to describing how the resulting techniques have been integrated and applied to a proof of concept smart home scenario. The resulting system is capable of providing useful services and pro-actively assisting the users in their activities. RUBICON learns through an incremental and progressive approach driven by the feedback received from its own activities and from the user, while also self-organizing the manner in which it uses available sensors, actuators and other functional components in the process. This paper summarizes some of the lessons learned by adopting such an approach and outlines promising directions for future work.