Design of Interface Module on Service Robot for Identification of Human Voice Direction

A domestic service robot (DSR) must have the ability to communicate with humans, such as understanding human instructions and responding appropriately, to interact with them. In this study, an interface module is developed to facilitate the human-robot interaction. The module consists of a camera that acts as an image processing sensor, a microphone array to collect the human voice, a speaker to output the robot voice, and an embedded computer to process the data. In this paper, we assign a task to conduct the interaction between humans and robots using the developed interface module. The assigned task is to detect the source of the sound and move in the direction of the detected sound and validate whether the sound is from a human or not. To do that, first, we estimate the direction of the sound collected from a microphone array. Next, a servo motor controlled by an Arduino Nano is used to direct the robot according to the sound coming. Then, face detection is performed on captured images from the camera to finalize the direction as well as check the human. Thanks to the servo motor, we can move the robot to the final position. To evaluate the interface module, we conducted experiments in an empty room-sized $2 \times 2 m^{2}$. The distance between the sound source and the developed module is one meter. Here, the accuracy for speech recognition and the direction of the sound were 85.34 %, 86.66%, respectively. The face detection required a minimum distance of 20 cm from the camera. Moreover, we discovered that the resolution of the coverage direction of the sound is 45 degrees.