An architecture that incorporates a seamless integration of different leaming paradigms is presented.Sensor processing, recurrent neural networks, learning from experience ana' qualitative knowledge are the key elements of the system. The goal applications are those tasks which cannot be fully programmed due to uncertainties and incomplete knowledge. The proposed sensor-based architecfure combines several leaming paradigms as well as pre-programmed modules, since experimental evidence suggests that some paradigms aye more convenient for leaming certain skills. The correspondence between qualitative states and actions is learnt. The qualitative treatment of information makes it suitable for the analysis of system behavior, knowledge extraction and generalization to other more complex tasks. Programming is used to decrease the complexity of the learning process. This general approach is a suitable scheme for a wide range of robot situations. Results wz provided for the simulation of a sensor-based goallfinding task as well as f o r a real application of the architecture in a robotic insertion process in three dimensions.