Diseño e implementación de un sistema de visión artificial mediante una cámara RGB-D para la interpretación de señalización industrial en un robot bípedo.
Loading...
Date
2026
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad de Concepción
Abstract
Esta memoria de título abarca el desarrollo del diseño e implementación de un sistema de visión artificial mediante una cámara integrada en un robot bípedo, con la finalidad de que pueda identificar e interpretar señaléticas industriales. El principal propósito fue proporcionar al robot de capacidades visuales básicas que le permitan poder percibir su entorno y por medio de ello lograr responder a estímulos relevantes, que pueden ser, la identificación objetos por medio de su forma o colores específicos. El sistema fue diseñado empleando una cámara Intel RealSense D435i, la cual fue integrada a una plataforma de procesamiento basada en Raspberry Pi 5, mediante el uso de la biblioteca OpenCV, donde se utilizaron técnicas de procesamiento de imágenes, logrando la capacidad de detectar características visuales como formas y colores que se relacionan con la señalización de seguridad en entornos industriales. Además, la cámara proporciona información sobre la profundidad, lo que permite al sistema calcular la distancia entre el robot y los objetos detectados, lo que a su vez facilita generar acciones de navegación en relación a la cercanía de la señal detectada, tales como reducir la velocidad, detenerse o cambiar de dirección. Se realizaron pruebas experimentales enfocadas a evaluar el funcionamiento del sistema ante diferentes escenarios de detección de señalización industrial. Los resultados adquiridos mostraron que el sistema puede identificar correctamente varias señales de seguridad y tomar decisiones dentro del rango operativo del sensor de la cámara.
This thesis covers the design and implementation of a machine vision system using a camera integrated into a bipedal robot, with the aim of identifying and interpreting industrial signage. The main objective was to provide the robot with basic visual capabilities that allow it to perceive its environment and, through this, respond to relevant stimuli, such as identifying objects by their shape or specific colors. The system was designed using an Intel RealSense D435i camera, which was integrated into a Raspberry Pi 5 based processing platform using the OpenCV library. Image processing techniques were employed, achieving the ability to detect visual characteristics such as shapes and colors related to safety signage in industrial environments. Furthermore, the camera provides depth information, allowing the system to calculate the distance between the robot and the detected objects. This, in turn, facilitates navigation actions based on the proximity of the detected signal, such as reducing speed, stopping, or changing direction. Experimental tests were conducted to evaluate the system`s performance under different industrial signal detection scenarios. The results showed that the system can correctly identify various safety signals and make decisions within the camera sensor`s operating range
This thesis covers the design and implementation of a machine vision system using a camera integrated into a bipedal robot, with the aim of identifying and interpreting industrial signage. The main objective was to provide the robot with basic visual capabilities that allow it to perceive its environment and, through this, respond to relevant stimuli, such as identifying objects by their shape or specific colors. The system was designed using an Intel RealSense D435i camera, which was integrated into a Raspberry Pi 5 based processing platform using the OpenCV library. Image processing techniques were employed, achieving the ability to detect visual characteristics such as shapes and colors related to safety signage in industrial environments. Furthermore, the camera provides depth information, allowing the system to calculate the distance between the robot and the detected objects. This, in turn, facilitates navigation actions based on the proximity of the detected signal, such as reducing speed, stopping, or changing direction. Experimental tests were conducted to evaluate the system`s performance under different industrial signal detection scenarios. The results showed that the system can correctly identify various safety signals and make decisions within the camera sensor`s operating range
Description
Tesis presentada para optar al título de Ingeniero/a Civil Eléctrico/a.
Keywords
Visión artificial, Robótica Chile, Sensores