Human-Drone Interaction: Using Pointing Gesture to Define a Target Object

This paper focuses on exploring the optimal gesture interface for Human-Drone Interaction in a firefighting scenario. For this purpose, we conducted a Preliminary Interview and User Study with seven subjects from the Kobe Firefighting Brigade, Japan. As a drone's flight and locomotion properties significantly affect the user's mental and physical expectations, differently compared to other grounded robots, a careful investigation of user-defined design preferences and interactions is required. This work proposes an examination and discussion, with experienced firefighters, about Human-Drone Interactions when relying solely on the drone's monocular camera, without relying on other devices such as GPS or external cameras. The User Study had three main elements: A drone, a building, and the volunteering firefighters. During the Study, each firefighter should specify a window to the drone. The drone would theoretically use that window to enter the building and perform some designed tasks, like information gathering, thus saving the firefighter time and effort. Results show that the subjects always chose Pointing Gestures and voice commands as a means to communicate to the drone a target window. We built A prototype application with the resulting gesture from this investigation. In the prototype, a drone can understand to which object the user is pointing.