DESIGN AND IMPLEMENTATION OF VISUAL BASED TRACKING SYSTEM FOR LUMEN AS WAITER ROBOT

• Main Author: NALA HUSNA NIM : 13212071, DEWI
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/21689
• Institution: Institut Teknologi Bandung

Abstract - Lumen Social Robot is a humanoid robot development in order to make robot to be a friend to human. In this study, Lumen developed into a waiter. The role as a waiter is limited to deliver a menu to customers, as well as perform simple interactions with customers. In order to carry out the role, Lumen is supported by Nao robot, a server, and several processing modules. <br /> <br /> <br /> One of the necessary module is tracking module. This work will talk about design and implementation of a module to estimate the position of customer based on visual input which is camera of Nao Robot, then called Visual Based Tracking. It will compute customer’s positions : distance and angle, then the information could be used to move the robot toward the object. Visual Based Tracking module will use width and center of frame of face detection rectangle to compute the information needed. Determination of the angle is implemented by using the comparison between the distance object center to center of the camera to width of whole image, then connected to the Horizontal Field of View (HFOV) for horizontal angle or Vertical Field of View (VFOV) for vertical angle. Distance estimation is done by using triangle similarity by utilizing the width of the face detection. <br /> <br /> <br /> Accuration problem arise when we use frame of face detected rectangle to get distance of customer so another method is developed. The alternative is using interpupillary distance from image to compute distance of customer. Through experiment, it is clear that using frame detection of face is much better than using interpupillary distance (eye detection).