The main goal of the project is the development of generic robot software components for robotic manipulation as well as the learning of new situation-specific behaviours both for its use in robotic manipulators of different morphology. A set of generic behaviour templates are stored on the robot for solving specific parts of a complex manipulation task. The first part of the talk will talk about the current status of the development of the robot software components required to make use of those behaviour templates and to ensure a successful manipulation by incorporating sensor-based reactive strategies which take into account whole body robot dynamics and kinematics. In case of a missing behaviour template for a certain subtask, the robotic system incorporates an interface to a human operator, who via demonstration will show the robot how to deal with the new situation and will generate a new behaviour template to be stored on the robot for its future retrieval and use. The second part of the talk will emphasize on the pipeline for imitation learning, from the operator's motion capturing, to the re-inforcement learning techniques through the automatic behaviour segmentation of observed movements.