3D visibility check in webots for human perspective taking in human-robot interaction

The rapid development of intelligent robotics would facilitate humans and robots will live and work together at a human workspace in the near future. It means research on effective human-robot interaction is essential for future robotics. The most common situation of human-robot interaction is that humans and robots work cooperatively, and robots should give proper assistance to humans for achieving a goal. In the workspace there are several objects including tools and a robot should identify the human intended objects or tools. There might be situational differences between a robot’s perspective and a human perspective because of several obstacles in environment. Thus, a robot needs to take the human perspective and simulates the situation from the human perspective to identify the human intended object. For human perspective taking, first of all a robot needs to check its own visibility for the environment. To address this challenge, this paper develops a 3D visibility check method by using a depth image in Webots. By using the developed method, a robot can determine whether each point in the environment is visible or invisible at its posture and detect objects if they are visible.