ABSTRAK Chandra Liuswanto
PUBLIC Alice Diniarti COVER Chandra Liuswanto
PUBLIC Alice Diniarti BAB 1 Chandra Liuswanto
PUBLIC Alice Diniarti BAB 2 Chandra Liuswanto
PUBLIC Alice Diniarti BAB 3 Chandra Liuswanto
PUBLIC Alice Diniarti BAB 4 Chandra Liuswanto
PUBLIC Alice Diniarti BAB 5 Chandra Liuswanto
PUBLIC Alice Diniarti BAB 6 Chandra Liuswanto
PUBLIC Alice Diniarti PUSTAKA Chandra Liuswanto
PUBLIC Alice Diniarti
A positioning system is one of the most critical fragments to develop a fully autonomous
vehicle. Generally, any Unmanned Aerial Vehicle (UAV) or drone, including a multirotor, is
paired Global Navigation Satellite System (GNSS) with an inertial sensor as the primary
positioning system. The inertial sensor has a drifting issue and can only provide short-term
navigation; likewise, the inertial sensor needs to be corrected by an absolute positioning system
such as GNSS. Under an indoor environment, the GNSS positioning system is unreliable
because the signal will be prone to multipath error and signal degradation due to being blocked
and reflected by the wall. An alternative positioning system in an indoor environment is
required to develop a navigation system for an indoor multirotor.
In this research, a positioning system, a stereo camera system, is developed based on videobased
motion capture to estimate the position of a micro-multirotor in an indoor environment.
The system consists of identical two digital cameras separated by a baseline and able to capture
images simultaneously. Afterward, the images are segmented to estimate the center of the
interest object using a color segmentation and the object's position is calculated based on the
triangulation of the center of the interest object.
The system is tested with multiple objects, from a digital image, a pendulum, and a microquadrotor.
The result shows that the stereo camera system has high precision and capable to
output position data that can be used for navigation data. The stereo camera system performance
is categories into three categories based on precision of the stereo camera system to estimate
position of the object with dependent on the object’s velocity and depth of the object. While
testing on a proof of concept device, a micro-quadrotor, the output has high frequency noise.
The output can be filtered using a moving average filter and a proposed Kalman filter model to
reduce the noise in the stereo camera system.