We present a framework for the development of robots able to learn through dialogue. We starts by presenting the motivation, problems and objectives of the work, as well as some related work in the different areas our project covers which might be useful when developing this kind of robots. Dialogue systems, intelligent agents and knowledge representation are some of the areas covered. The framework’s has a functional, layered architecture, with the mind on the top layer, competences on the middle layer and the body on the bottom layer. This architecture focuses on extensibility, flexibility and portability. We present a test scenario and a deployment of the architecture for that scenario: the a robot is able to learn different colors and identify them later.