This paper presents a discussion about the implementation of mechanisms of fuzzy logic for recognizing patterns and parameters in order to predicting the movement intention from a user, through muscle activity in upper and lower members in adult human beings. This information is obtained through reading surface electromyography and is debugged and analyzed by a fuzzy logic system calculated for the prediction of intended movement. A brief comparison between fuzzy sets of straight lines and curved lines is done to determine the best system in generating reliable control orders for prototype robotic exoskeleton type. The fuzzy system designed is Mandani type with nine rules in the inference engine, which had two stages of interaction with samples: one design and other validation, where sought meet an initial threshold of 80 % of effectiveness. Used analog and digital components for data acquisition processes to perform amplification, filtering, digitization and transmission of samples which were implemented in full. The fuzzy system has four input parameters easy obtaining the electromyographic signal input, looking it fast execution at a later real time application. As an important part harmonization by using RMS envelope to make the system more robust against disturbances in the samples. This information is used for generate control commands to an exoskeleton type robotic system to support some user activities.