Date of this Version
Seth Doebbeling. Cyber-physical system characterization and co-regulation of a quadrotor uas. Master's thesis, University of Nebraska-Lincoln, 2017.
An Unmanned Aircraft System (UAS) is a Cyber-Physical System (CPS) in which a host of real-time computational tasks contending for shared resources must be cooperatively managed to obtain mission objectives. Traditionally, control of the UAS is designed assuming a fixed, high sampling rate in order to maintain reliable performance and margins of stability. But emerging methods challenge this design by dynamically allocating resources to computational tasks, thereby affecting control and mission performance. To apply these emerging strategies, a characterization and understanding of the effects of timing on control and trajectory following performance is required. Going beyond traditional control evaluation techniques, in this work we characterize the trajectory following performance, timing, and control of a quadrotor UAS under Discrete Linear Quadratic Regulator control (DLQR) designed at various sampling rates. We introduce new metrics for characterizing cyber-physical quadrotor performance, and provide empirical evidence that high-samplingrate control strategies over 50 Hz may not significantly improve control performance for our quadrotor platform and hence may not effectively allocate resources that could be used to improve other (non-control related) mission objectives. We then propose a strategy in which a model representing the sampling rate is augmented to the state-space model of a quadrotor UAS, controllers are designed for this holistic system to more effectively allocate these resources. We develop a full nonlinear equation co-regulation simulation suite in MATLAB and provide analysis of the UAS in following a trajectory under traditional control design as well as our proposed co-regulation design for comparison. Under co-regulation we are able to reduce the maximum power consumption by ~12% and the time averaged normalized state error by ~75% for a unit step in x, y, and z, while maintaining relatively good cross-tracking performance. Results illustrate the need for a higher level trajectory planning and generation technique capable of translating mission tasks into smooth trajectories and providing both physical and cyber reference commands suitable for our variable rate co-regulation architecture. Therefore, we design a low level, kinematic trajectory generator capable of easily adjusting timing constraints which provides a first step toward such a motion tracking architecture.
Adviser: Justin Bradley
Acoustics, Dynamics, and Controls Commons, Navigation, Guidance, Control and Dynamics Commons, Robotics Commons, Systems Engineering Commons, Systems Engineering and Multidisciplinary Design Optimization Commons