Human-face-tracking using visual servoing by patient robot

Human-body models imitating a part of human body have been used for medical training. But those models seem not to suit training because they cannot be felt to be similar to actual humans. Therefore, the simulator being felt like a real human is required for the medical trainees being able to assess patient's physical and emotional conditions, since medical care procedures are vital to watch out patient's pains given by the care treatment itself to avoid failures leading the patients to dangerous conditions. Based on the needs for medical training by using human-like robot, we have developed a simulator for injection training called "patient robot." Patient robot has to behave like humans to make the medical training effective, so we introduced a visual servoing system into the patient robot so that the patient robot can gaze at a trainee's face like humans or turn away its face form the medical procedures. We executed some experiments of gazing at a rotating object using visual servoing, in which actual object's positions, patient robot's gazing point and pose-tracking data have been measured. We confirmed patient robot can track real human faces.