Disturbance effects acting on small fixed-wing UAVs should be explicitly considered and ultimately eliminated. This paper proposes a post-controller disturbance observer-based nonlinear path for a small fixed-wing UAV affected by disturbances such as wind.
Motivation
Research Objectives
A method that uses ground-based measurements on trajectory tracking is further explored using multiple UAVs [10]. In addition, DOBC was used in the longitudinal controller of a small fixed-wing UAV affected by disturbances such as wind [16], which was further investigated by using DOBC with a linear quadratic controller (LQR) to improve its performance [17] .
Outline of the Thesis
It is important that the UAVs autonomously follow a predetermined path to perform the mission, such as surveillance and search. There are many available path tracking methods, it is important for UAVs to follow the path precisely and robustly to perform its mission.
Path Following Problem
Path tracking is generally required for small fixed-wing UAVs and any type of unmanned vehicle. The most common missions for UAVs are to follow either straight lines or circular orbits.
Carrot-Chasing
Pure Pursuit and LOS
Linear Quadratic Regulator
Nonlinear Guidance Law
Vector Field
Disturbances are widely found in control systems and adversely affect the performance or stability of the control systems. The disturbance is difficult to measure directly using the sensor, and disturbance rejection is one of the main challenges in control system design. Disturbance observer-based control technique has been considered as a popular method that estimates and compensates for the disturbance.
In the case of existing robust control techniques, a feedback technique is basically used to reject disturbances. To implement feedforward control, it is necessary to measure the disturbance acting on the system. A disturbance observer that estimates disturbances using the concept of an observer to estimate system disturbances and uncertainties was studied, and a control scheme based on a disturbance observer that compensates for disturbances using the estimated disturbance was studied.
Basic Framework
Disturbance rejection is performed based on errors between the output state and the desired command [16]. Compared to feedback control, their methods produce a slow response in reducing disturbance effects, since feedback control is performed based on tracking errors. Compared to PADC techniques, which only reject disturbances with a passive feedback rule, disturbance observer-based control (DOBC) always leads to a faster dynamic response when dealing with disturbances, as a compensation term is provided by providing an intervention to offset the disturbances directly. [28].
DOBC achieves the detection and disturbance rejection performance by using the feedback and feedback controllers. This means that DOBC is glued to the designed controller to improve the ability to reject disturbance and improve the robustness. The DOBC technique has been widely used and applied to many industrial systems, robotics, flight control, and aerospace systems [ 28 – 30 ].
Frequency Domain Disturbance Observer
The nominal system Pn(s) can be defined as:. where u¯ is the external input of the attitude controller. The desired control input can be decided such that the system of Eq. 1) with disturbances becomes equivalent to the nominal system of Eq. However, the use of the desired control input is not directly feasible since the f(x), g and dof actual system are unknown.
The frequency domain disturbance observer wants to calculate the position controller output yp and the system input. Q-filters, which are considered stable low-pass filters, should have relatively more degrees than the nominal device as:.
Time Domain Disturbance Observer
Taking the derivative of Eq. Provided that the disturbances tend to be constant, this means that disturbances vary slowly, and the gainL is appropriately determined so that-LBdis Hurwitz, the Eq. 10) is shown that perturbation estimation error is asymptotically stable.
Nonlinear Disturbance Observer
The outer loop controller, including the heading, altitude and speed controller, is designed as a proportional-integral (PI) control method with overshoot protection, which works to relieve the integrator build-up and allow the UAV to follow the heading, altitude and speed command. The Lyapun Vector Field (LGVF) trajectory tracking technique is used to track the reference circular path for UAVs, which has been mainly studied in [7–9] for UAV tracking without considering wind disturbances. The Nonlinear Disturbance Observer Based Control (NDOBC) method used by Liu et al in [18] is used for precise trajectory control for UAVs affected by disturbances such as wind.
The disturbance estimated by the nonlinear disturbance observer (NDO) is incorporated into the controller following the LGVF to compensate for the disturbance effects.
Outer Loop Controller Design
9 is used to relieve integrator stacking in the outer loop controller with PI control. Anti-termination schemes are designed to limit the termination of the integrator after u, such as φcmd, θcmd and tcmd is saturation.
Unmanned Aerial Vehicle Kinematics
Nonlinear Disturbance Observer Design
If NDO gainl(x) is correctly chosen, Eq. 25) can be asymptotically stable regardless of the state x.
Lyapunov Guidance Vector Field
The wind disturbance is estimated by a nonlinear disturbance observer (NDO) in Eq. 23) and the target is locked, the guidance vector is adjusted to take into account the speed [Tx, Ty] = [Wx, Wy]. To compensate for the perturbation estimated from the nonlinear perturbation observer, the new desired speed of LGVF is used for Eq. 33) where Wˆx,Wˆy is the perturbation estimated from Eq. 23), and is a scale factor that can be obtained by equating the magnitude of the original and the new desired speed.
![Figure 10: The geometry of tangent vector field [8].](https://thumb-ap.123doks.com/thumbv2/123dokinfo/10510012.0/30.892.207.685.506.1015/figure-10-geometry-tangent-vector-field-8.webp)
The Proposed NDOBC-Based LGVF Controller
Numerical simulation is first performed through the MATLAB Simulink environment for proof of concept. MATLAB Simulink simulation using the LGVF and DOBC methods is performed using a two-dimensional kinematic model of the UAV instead of using a controller such as PI controller to easily verify the performance of the DOBC.
MATLAB Simulink Structure
Simulation Result
In this study, software in the loop simulation is performed using Gazebo simulator with PX4 open source autopilot software and robot operating system (ROS). The software in the loop simulation enables the development of algorithms that can be deployed in real UAVs with many instances.
Gazebo Simulator
PX4 Autopilot Software
The flight stack is the software of the flight control system and estimation, and the middleware is a robotic part that can support all kinds of UAVs and UGVs, providing communication and hardware integration. The upper part of the diagram contains the middleware block and the lower part is the part of. It aims to adjust the value of the output state to match the desired command.
For example, the position controller uses the desired position as a command, the output state is the currently estimated position by the estimator, and the output is the attitude and thrust command that moves the vehicle to the desired position by the attitude controller. A mixer mixes power commands coming from the mode controller and distributed into individual motor inputs. The attitude controller is tuned to the airspeed measurements that can be measured by the airspeed sensor due to high speed effect.
Robot Operating System
The forces and moments about the axis of the body of the aircraft are created by the control surfaces and the aerodynamic damping which interfere with the motion proportional to the speed of the body. Damping can be compensated for by using forward thrust in the velocity circuit in order to maintain a constant velocity. The middleware allows the PX4 autopilot software to run on the desktop operating system which can use ROS (Robot Operating System).
Robot Operating System with Gazebo Simulation
Robotics System Toolbox In Simulink
Software In The Loop Configuration
The outer loop controllers such as heading, altitude and speed controller are tuned during the simulation test. The desired mode, such as roll and pitch, is sent from the outer loop controller to the PX4 autopilot. In this study, the PX4 autopilot is used to follow the attitude command, such as desired roll and pitch, from the NDOBC-based LGVF controller.
MATLAB Simulink environment can communicate with the PX4-based Gazebo simulator using the ROS.
Simulation Result
After confirming the performance of the outer loop heading, altitude and speed controller in Fig. 24, 25 and 26, SITL simulation using NDOBC and LGVF methods is performed in an environment where there is no disturbance such as wind. It is shown that the circular pattern following the performance is almost the same as that of the LGVF without a nonlinear perturbation observer.
The advantage of DOBC is that the nominal performance of the designed controller is recovered in the absence of disturbances or uncertainties. It is shown that the UAV follows the command of height and speed well in Fig. Outdoor flight experiments are performed based on the software in the loop simulation algorithm that can be deployed in the small fixed-wing UAV in wind condition at about 3m/s.
Skywalker X-5 UAV
Auto Code Generation
C/C++ code so it can be installed on an embedded computer such as a Raspberry Pi that does not support Simulink models. A powerful feature of automatic code generation is that communication delays are greatly reduced by entering the code into the UAV computer. The proposed MATLAB Simulink environment algorithm is installed on a Raspberry Pi device, which is an embedded system with automatic code generation.
Flight Experiment Configuration
Experiment Result
34 shows that the X-5 UAV using LGVF without DOBC follows a path far away from the reference path due to wind effect. DOBC helps reject wind disturbances and results in accurate path tracking performance. This paper proposes an LGVF trajectory tracking controller based on a nonlinear disturbance observer for a small fixed-wing UAV affected by disturbances such as wind.
Software in cycle simulation is performed to verify the performance of the proposed algorithm which can be deployed on small fixed wing UAV. Real flight experiments using the X-5 UAV demonstrated the performance of the following proposed routing method. Observer-based disturbance control with contra-refraction applied to a small fixed-wing UAV for disturbance rejection.
