Automatic System Identification Control Methodologies for Autonomous Aerial Vehicles
thesis
posted on 2016-12-08, 01:00 authored by Man Ho ChoiThe
system identification methodologies of aerial vehicles can be developed by
experimental analysis, computational fluid dynamics analysis or heuristic
analysis. However, these methodologies require accurate estimation or
experimental tuning process before implementing to the autonomous control
methodologies. This research was undertaken to develop autonomous system
identification control methodologies suitable for a broad range of small-scale
aerial vehicles that do not require any prior knowledge of the aerial vehicle.
In order to develop the proposed autonomous system identification control methodologies for aerial vehicles, the first part of the thesis describes the sensing equipment, which collects all the crucial information, and the experimental platform. This research utilised low cost equipment for the small-scale aerial vehicles. Although the low cost equipment suffers from bias error drift, it has the advantage of reduced weight and ideal for the small-scale aerial vehicles.
The measurements from the sensing equipment cannot determine the required information by numerical integration because the low cost sensing equipment suffers from the bias error drift which mentioned above. The second part of this thesis presents the sensor fusion algorithms to overcome the bias error drift. The accelerometers, gyroscopes and magnetometers were integrated to estimate the attitude of the aerial vehicle. The global positioning system, accelerometers and gyroscopes are integrated to determine the position and velocity of the aerial vehicle.
The traditional autonomous control methodologies require the determination of aerodynamic derivatives of the aerial vehicle. Wind tunnel testing and computational fluid dynamics determination are commonly utilised to estimate the aerodynamics derivatives. However, these methods require accurate knowledge of the aerial vehicle prior to the autonomous flight. This thesis characterises the performance of system identification methodologies based on the flight history of the aerial vehicle.
The system identification results are further analysed and utilised to develop the autonomous control methodologies. However, the system identification results involve uncertainties. The robustness of the control methodologies is characterised in this research. Proportional-integral-derivative (PID) control method, optimal control method and sliding mode control method are computationally and experimentally investigated. The performance of these control methodologies is studied and presented by the results of the system identification methodologies. The computational and experimental results showed that the sliding mode control and optimal control methodologies were improved from PID control methodology.
In order to develop the proposed autonomous system identification control methodologies for aerial vehicles, the first part of the thesis describes the sensing equipment, which collects all the crucial information, and the experimental platform. This research utilised low cost equipment for the small-scale aerial vehicles. Although the low cost equipment suffers from bias error drift, it has the advantage of reduced weight and ideal for the small-scale aerial vehicles.
The measurements from the sensing equipment cannot determine the required information by numerical integration because the low cost sensing equipment suffers from the bias error drift which mentioned above. The second part of this thesis presents the sensor fusion algorithms to overcome the bias error drift. The accelerometers, gyroscopes and magnetometers were integrated to estimate the attitude of the aerial vehicle. The global positioning system, accelerometers and gyroscopes are integrated to determine the position and velocity of the aerial vehicle.
The traditional autonomous control methodologies require the determination of aerodynamic derivatives of the aerial vehicle. Wind tunnel testing and computational fluid dynamics determination are commonly utilised to estimate the aerodynamics derivatives. However, these methods require accurate knowledge of the aerial vehicle prior to the autonomous flight. This thesis characterises the performance of system identification methodologies based on the flight history of the aerial vehicle.
The system identification results are further analysed and utilised to develop the autonomous control methodologies. However, the system identification results involve uncertainties. The robustness of the control methodologies is characterised in this research. Proportional-integral-derivative (PID) control method, optimal control method and sliding mode control method are computationally and experimentally investigated. The performance of these control methodologies is studied and presented by the results of the system identification methodologies. The computational and experimental results showed that the sliding mode control and optimal control methodologies were improved from PID control methodology.
