A Study on Motion Control of Gimbal-based Target Tracking System

Abstract

This dissertation pursues solutions to the problem of controlling a camera’s posture with a target tracking system using a two-axis gimbal mechanism. Carried on a mobile vehicle, the camera’s line-of-sight (LOS) needs isolating from the vehicle’s motions and disturbances, and at the same time, it rotates to follow a predefined target. Besides the external disturbances, the control objective is also challenged by the coupled kinematics and nonlinear dynamics of the gimbal itself. Moreover, the use of low-cost devices causes fundamental limitations, typically time delay and unreliable measurements, to the control system. The mathematical model formulated in this dissertation investigates the effects of these factors on the characteristics of the controlled system. Thereby, the design of the controllers takes them into account such that the control systems not only fulfill the above-mentioned objective but also achieve the following desired features. Firstly, visual servoing is attained using the camera available in the system, which usually serves only as a payload, to detect and locate a target of interest. A proposed control configuration consists of an image tracker implementing the Kernel Correlation Filter (KCF) algorithm, a calculation of the gimbal’s desired motion considering the coupling influence, and a control law derived by the backstepping technique. In this task, the camera is required to be stable, and the target’s projection is brought to the center of the image plane. The kinematic coupling and imaging geometry are the main challenges. The proposed vision-based control system decouples both couplings simultaneously, so conventional control methods can be easily implemented, and good tracking performances are obtained. Secondly, accurate and real-time tracking of an arbitrary reference with the gimbal system is considered. For this objective, the control systems must overcome the difficulties, namely the gimbal complexities, coupled measurements, and especially, the delay time constraints. Two novel nonlinear controllers are discussed, one is based on the sliding mode control with the super-twisting algorithm (STSMC), and the other is a nonlinear backstepping control with the time-delay compensation (TDC-Backstepping). The design of the former treats all non-ideal factors as matched and unmatched disturbances. The super-twisting algorithm with a chattering-reduced signum-like function suppresses the oscillations of the delayed system while still ensures the boundedness of the control error. On the other hand, the recursive design procedure of the latter deals with each factor in a corresponding step. The result is a memory-based control law that enhances the tracking performances and expands the system’s bandwidth. Finally, a robust fault-tolerant control (FTC) scheme provides simultaneously reliable tracking and effective disturbance rejection, even when a sensor’s fault occurs and severe disturbances affect the system. Representation of the faulty gimbal system with its mentioned constraints and disturbances is introduced. The designed controller consists of two components: (a) An unknown input observer (UIO) acquires information on the fault and disturbances, whose efficiency is proved using the linear matrix inequality (LMI) technique; and (b), a robust control law, using the observer estimations, is obtained based on the combination of the super-twisting algorithm, backstepping procedure, and integral sliding mode control (SMC) technique. Moreover, for each design of the control systems, comparative simulations and experimental studies are conducted, and their results validate the efficiency of the designed systems. The solutions obtained in this dissertation not only apply to the specific apparatus but also can extend to other subjects and mechanisms.