Dynamic Safety Zones for Industrial Robots: A Fuzzy Logic and Computer Vision Approach

• Lucas C. Sousa CEFET-RJ
• Vinícius B. Schettino CEFET-MG
• Murillo F. Santos CEFET-MG
• Tatiana M. B. Santos UFF
• Diego Haddad CEFET-RJ
• Milena F. Pinto CEFET-RJ

Resumo

This research focuses on integrating a robotic manipulator in shared workspaces within the framework of Indus-try 4.0 for Human-Robot Collaboration (HRC). The primary objective is to implement a safe strategy by employing a fuzzy-speed controller to enhance the mobile robot's movements near machinery. This strategy utilizes two cameras: one mounted on the ceiling of the industrial environment and another on the robot itself, called an eye-in-hand camera. The You Only Look Once Version 3 (YOLOv3) Convolutional Neural Network (CNN) detects the obstacles in the environment and identifies the target objects. Simulations were conducted using Gazebo software along with a Robot Operating System (ROS) to determine the effectiveness of the proposed approach in ensuring safe robot movements while accurately reaching target positions by dynamically adjusting velocity within the defined danger zone.

Palavras-chave: YOLO, Service robots, Operating systems, Robot vision systems, Cameras, Software, Safety, Convolutional neural networks, Object recognition, Robots, Shareable Workspace, Autonomous Robot, Industry Environment