Target Tracking Of Quad-Rotor UAV Using Adaptive Sliding Mode Controller Based on Real Time Image Image-Processing
No Thumbnail Available
Date
2023-03
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
Surveillance plays a crucial role in various military and civilian operations, including search and
rescue missions. In recent times, unmanned aerial vehicles (UAVs) have gained significant
popularity and are considered the ideal resources for such applications.
The quad-rotor offers distinct advantages over fixed-wing light vehicles, including costeffectiveness,
compact size, and the ability to hover, as well as perform vertical takeoff and
landing. However, to ensure reliable performance in various tasks, it is crucial to have a
specialized controller that can effectively account for the quad-rotor's nonlinear dynamics, underactuated
characteristics, and uncertainties related to parameters and external disturbances.
To address these challenges and enhance controller robustness, a slide mode control technique is
employed, which offers advantages over traditional PID and other nonlinear controllers. This
control design also incorporates image processing capabilities to enable real-time identification
and tracking of user-defined targets, providing efficient and accurate performance.
This thesis aims to develop real-time target identification and tracking system based on image
processing. The system utilizes fast decision-making capabilities and robust flight control
techniques to ensure optimal trajectory for a quad-rotor. The image processing component relies
on an onboard camera and employs the You Only Look Once (YOLO) algorithm to identify and
estimate the continuous motion coordinates of the target. The identification process primarily
relies on user-defined criteria such as shape, size, and color. The graphics processor of the
embedded software is responsible for accurately calculating the target dynamics relative to a
common reference frame of Earth's geographic coordinates. Furthermore, the system selects
efficient maneuvers based on time and energy considerations, leveraging the capabilities of the
YOLO algorithm.
In the presence of parametric uncertainties and external disturbances, the controller effectively
minimizes tracking errors within a short time frame, ensuring obstacle clearance and reducing
redundancy costs. Through simulation results, the designed controller demonstrates minimal
altitude and attitude tracking errors, achieving precise identification and tracking of a userdefined
ground target.