Deep Learning Model for the Recognition of Its Environment of an Intelligent System

This research project consisted in the design of an Artificial Neural Network with Deep Learning for an intelligent system, they were installed twelve sensors ultrasonic HC-SR04, which ones detected and learned all kinds of obstacles, such as: walls, tables, chairs and others. The methodology used was the concurrent design has five phases: the first the conceptual plan was carried out, the second a kinematic study, the third a dynamic study, the fourth a mechanical project and finally the simulation of the system. Artificial Neural Networks were designed with Deep Learning and trained with the Backpropagation algorithm. The ANN was programmed and recorded in the Arduino Mega 2560 module. All the corresponding simulations were carried out, it was verified that the ultrasonic sensors have sent the signal to the Artificial Neural Network with deep learning and they carried out a learning of their environment, they were also checked the displacement of the mobile robot resulting in the desired performance. In conclusion, the proposed design was achieved and it was simulated with all kinds of events, in addition it was verified that Artificial Neural Networks with deep learning can detect and learn from all the obstacles that are in their environment.