The combination of virtual reality interactive systems and educational technologies have been used in the training of procedural tasks, but there is a lack of research with regard to providing specific assistance for acquiring motor skills. In this paper we present a novel approach to evaluating motor skills with an interactive intelligent learning system based on the ULISES framework. We describe the implementation of the different layers that ULISES is composed of in order to generate a diagnosis of trainees' motor skills. This diagnostic process takes into account the following characteristics of movement: coordination, poses, movement trajectories and the procedure followed in a sequence of movements. In order to validate our work we generated a model for the diagnosis of tennis-related motor skills and we conducted an experiment in which we interpreted and diagnosed tennis serves of several subjects and which shows promising results.