Implementación de un prototipo de sistema electrónico en un andador para niños que controle el acceso a lugares inseguros, usando Visión Artificial.

Abstract

In the present work of titration a prototype of system in a walker for children was implemented that controls the access to insecure places using atrificial vision. The system consists of five stages: training, information acquisition, processing, control and feeding. In the firt stage the training of the classifiers was carried out using the Train Cascade technique to obtain unique characteristics of the objects under study (outlet and harrows), in the second stage a Raspberry Pi3B processor was used, which receives the video in real time recorded by a Logitech C270 camera and the Arduino Mega 2560 microprocessor that receives the data sent by the distance sensor HC-SR04, the third stage is in charge of processing the video looking for characteristics of the objects obtained in the training stage, once the dangerous object (s) has has been identified, the control stage enters into operation, in wich the Arduino sends signals to the actuators: servo motor MG996R to prevent the movement of the walker and the Buzzer in charge of issuing an alarm signal that alerts the adult responsible for the child. Finally, the fifth stage feeds the system with an RPI Powerpack V2.0 battery of 3.8A and Lipo battery of two cells to 1A that give autonomy to the prototype. With the data obtained from the tests carried out, it was determined that the optimal distance and the recognition time of the objects is from 31 to 120cm and 3.8 milliseconds respectively, the data transmisión time is of 4 microseconds, this being an optimal time for the actuators work inmediately, it was also determined that the autonomy of the prtotype is 2 hours with 6 minutes, fulfilling the established requirements