Collaborative Robotics: Human-Robot Interaction through Gaze Tracking

Collaborative robotics is an increasingly relevant field in industrial automation, that aims to improve close interaction between human operators and robots. The main goal of this thesis is to explore human-robot interaction through gaze tracking, by using a UR5 robot from Universal Robots with 6 degrees of freedom and the free toolkit known as OpenFace 2.0. The project was inspired by the previous handover work implemented by the DIMEAS group at the Polytechnic University of Turin with the UR3 but with a different approach, focusing on visual tracking rather than physical handover. The OpenFace 2.0 toolkit played a key role in this work, due to its remarkable ability to extract facial parameters and gaze-related data. The gaze information extracted by OpenFace 2.0 were utilized to drive the robot end-effector towards the human’s point of focus in real-time. The implementation process encompasses control algorithms, system calibration and the integration of gaze tracking software with the UR5. In order to assess the effectiveness of this implementation, preliminary testing phases were conducted in a virtual environment using the CoppeliaSim software. Subsequently, physical tests were performed in the laboratory with the UR5 robot to verify the ability of the system to track eye movements and to respond appropriately.