Towards safe collaborative interaction empowered by face recognition

This paper presents the implementation of a face recognition node for a Sawyer collaborative robot using Robot Operating System (ROS), and its preliminary experimental validation. The node acquires images through the head camera of a Sawyer cobot and elaborates it using OpenCV libraries, applying the Viola-Jones algorithm to detect faces. The aim is to use face recognition functionalities to make the robot start its operations only when it correctly recognizes an operator in its workspace, thus increasing the safety of the human-robot interaction. The developed ROS node recognizes multiple faces at the same time and publishes the information required on a dedicated topic. The validation performed analyzed the success rate of the procedure at different orientations of the user's face.