UNMANNED AERIAL VEHICLE CONTROLLER DESIGN WITH VIRTUAL SENSOR BASED ON STOCHASTIC FRACTAL SEARCH OPTIMIZATION METHOD

• Main Author: Gusti Ngurah Agung Indra Man, I
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/41461
• Institution: Institut Teknologi Bandung

So far, Unmanned Aerial Vehicles (UAVs), especially quadrotor, are more widely used in outdoors, but along with the development, the use of quadrotors in indoors is increasing. Position measurement becomes a problem when using a quadrotor in the room. This research proposes a virtual sensor algorithm to estimate the position of the quadrotor without using physical sensors. The virtual sensor algorithm used consists of Diagonal Recurrent Neural Network (DRNN) and Unscented Kalman Filter (UKF). This research uses the quadrotor Parrot AR Drone 2.0. The Quadrotor Parrot AR Drone 2.0 has the internal algorithm for the speed control. For this reason, the dynamics modelling of quadrotor AR Drone 2.0 starts by approaching the internal algorithm. This approach is done by using two degree of freedom Proportional-Integral-Derivative (PID) controller. After the dynamics of systems is obtained, the quadrotor position controller is designed by using PID and Fuzzy Logic. The process of finding parameters in the design of virtual sensor, AR Drone 2.0’s dynamics modelling, and design of position controller are carried out using Stochastic Fractal Search (SFS) optimization algorithm. The SFS algorithm is used to minimize the likelihood of parameters trapped in local minimum values. After the integration of position controller and virtual sensor, a stable system response was obtained and the system was able to follow the various setpoints well.