Escalonador inteligente de tarefas para aplicações robóticas.

Abstract

This dissertation presents a Real-time Intelligent Tasks Scheduler developed to robotics application that implement microcomputer control. The proposed scheduler was idealized after observations realized in robots' control system utilized in Intelligent Automation and Neural Networks Laboratory (NEUROLAB) linked to Coordination of Computer Science Graduation (COPIN) of Federal University of Paraiba (UFPB). In these systems, each control stage was executed through real-time tasks. Several tasks were ordered by an expert and executed sequentially to a robot executed all necessary motions to realize an established goal. Was observed a constant interference of expert in ordination and time definition of each task until was to reach an optimal sequence of tasks to robot's control. With the Intelligent Tasks Scheduler, that uses rules and inference fuzzy, the ordination and time definition of each task now is realized by own system, permitting more autonomy to the system and consequentially becomes more intelligent.