The proposal of this work of Scientific Initiation consists in the implementation of a navigation system of an autonomous mobile robot that uses a controller by Artificial Neural Networks in a virtual environment. The control of the actuators of the virtual mobile robot, is done by an algorithm using Artificial Neural Networks programmed in Graphical User Interface (GUI). The information that feed the Neural Network is given by the sensors of the robot. The model used in the work comes from a real prototype consisting of two side wheels, each coupled to a dc motor, a silly wheel in the front and three ultrasonic distance sensors that assist the robot in the diversion of obstacles. The virtual model must move around and avoid obstacles created in the virtual environment using Neural Network properly trained with a database containing the information of distances of the sensors of the real robot, whose result must be visualized on the screen of a computer.
News published in Agência FAPESP Newsletter about the scholarship: