Human-Robot Interaction and Collaboration (HRI-C) Utilizing Top-View RGB-D Camera System

TARIQ TASHTOUSH; LUIS GARCIA; GERARDO LANDA; FERNANDO AMOR; AGUSTIN NICOLAS LABORDE; DAMIAN OLIVA; FELIX SAFAR

International Journal of Advanced Computer Science and Applications (IJACSA)

The Science and Information Organization

Año: 2021 vol. 12 p. 11 - 16

2158-107X

In this study, a smart and affordable system was developed utilizing an RGB-D camera to accurately measure the operator´s position relative to an adjacent robotic manipulator. This technology was implemented in a simulated human operation within an automated manufacturing robot, aiming to achieve two goals: enhancing safety measures around the robot by incorporating an affordable smart system for human detection and robot control, and developing a system that enables seamless human-robot collaboration to accomplish predefined tasks.The system employed an Xbox Kinect V2 sensor/camera and Scorbot ER-V Plus to model and simulate the selected applications. To accomplish the set goals, a geometric model was developed for the Scorbot and Xbox Kinect V2, robotics joint calibration was applied, an algorithm for background segmentation was utilized to detect the operator, and a dynamic binary mask was implemented for the robot. The efficiency of both systems was analyzed based on response time and localization error.The first application of the Add-on Safety Device aimed to monitor the workspace and control the robot to prevent collisions when an operator enters or approaches. This application aimed to reduce and eliminate physical barriers around the robots, expand the physical work area, minimize proximity limitations, and enhance human-robot interaction (HRI) in an industrial environment while maintaining a low cost. The system demonstrated an average response time of 500 ms to human intrusion, preventing collisions effectively. The system´s bottleneck was identified as the speed of communication between the PC and the robot.The second application involved developing a successful collaborative scenario between a robot and a human operator, where the robot would deposit an object onto the operator´s hand, simulating real-life human-robot collaboration (HRC) tasks. The system accurately detected the operator´s hand and its location, commanding the robot to place the object onto the hand. The system achieved a mean error of 2.4 cm in object placement. The limitation of this system was the internal variables and data transmission speed between the robot controller and the main computer.These results are encouraging, and ongoing work aims to experiment with different operations and implement real-time gesture detection in collaborative tasks, while ensuring the safety of the human operator and predicting their behavior.