#TITLE_ALTERNATIVE#

• Main Author: MUHAMAD LUTHFI IMAM NURHAKIM (NIM: 23609001); Advisor: Dr. Taufiq Mulyanto
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/16530
• Institution: Institut Teknologi Bandung

Optical measurement is an alternative method for indoor unmanned aerial vehicles (UAVs) pose estimation system. This type of measurement have several advantages such as the absence of noise generated from the accumulative errors caused by the integration as happened in inertial measurement, or electromagnetic noise appearing from direct current electric motor. However with limited sight distance and coverage area of observation, the optical measurement can be used as an alternative for indoor UAVs that present several advantages. <br /> <br /> <br /> <br /> This work presents development of an optical motion measurement system for 3D pose estimation in 6 degrees of freedom (DOF), especially for indoor UAVs, using dual cameras technique. The system consists of two outside-in virtual video cameras with 30 fps frame rate, an observe object model virtually with four spherical colored markers, a computing device and software to capture and process the digital image captured during motion. In this simulation, a virtual model with rectangular object with four different color markers attached at each corner will be used as observed object model. The whole system is done through computer simulation. The work starts from modeling an observe object model, then creating environment observation, capturing and processing the digital image, and finally reconstructing the position and attitude of an observed object model. Two virtual video cameras placed at different position capture every step of object movement. The recorded video is then separated into single digital image at each frame and processed to detect the centroid of ball marker using convert the RGB image to grayscale image then to binary. The information of 2D pair images from each single digital image is processed for reconstructing the position and orientation of the model. <br /> <br /> <br /> <br /> The calibration of virtual camera and reconstructing 6-DOF of the model used the Direct Linear Transformation (DLT) method. The simulation environment was carefully set-up to ensure that the image processing program could track the markers motion and the results could be calibrated properly. The system developed in this work is able to automatically track the markers movement. The pose estimation data obtained from simulation shows good precision with an acceptable error. For further development, this system might be able to used in flight testing of UAV.