Baldr Flight 11

Apurva, Brian, Chris, Danny, Julian, Laura, and Raghu arrived at the UMore Park Airfield around 8:30am for the eleventh flight of Baldr. In addition to Baldr flights, there were several Fenrir flights that are summarized in separate flight reports. Baldr is the UAV LabÕs newest UltraStick 120 airframe that will be used for aircraft reliability research. Baldr is a modified UltraStick 120 airframe that has split elevators and split rudders, each surface driven by a dedicated servo motor. Recently, efforts have been underway at the University of Minnesota to design fault tolerant control laws for UAVs. Specifically, researchers have been focusing on attempting to control Baldr using only the split elevators, with all other control surfaces locked into their respective trim positions. The key idea in this experiment is controlling a conventional aircraft with two coplanar control surfaces. There are two main motivations that drove this experiment: 1. Exploring the controllability of conventional aircraft (with an empennage) that have been severely handicapped with losses in multiple aerodynamic control channels, and 2. Drawing meaningful conclusions about the controllability of two-surface flying wing aircraft which are subject to faults in any one of the two aerodynamic control surfaces. For this experiment, the performance objectives were tracking phi and theta commands. Hence, only phi and theta tracking control loops were synthesized and implemented. It is important to note that each of the split elevators induce both longitudinal and lateral-directional motion in the aircraft. As a consequence, researchers were specifically interested in synthesizing multi-input, multi-output control laws (as opposed to the conventional loop-at-a-time designs). For this experiment, researchers synthesized a linear quadratic Gaussian (LQG) controller, with the primary performance objective being output regulation. A secondary performance objective was tracking phi and theta commands. In order to track commands, two integrators were added to the synthesized LQG controller on the roll and pitch channels. The integrators effectively ensure that the steady-state tracking error is as close to zero as possible. In addition, the baseline controller runs for the first 2 seconds before the LQG controller is engaged. This simulates a realistic scenario wherein the flight control law has to switch from the baseline to the backup after a fault has been detected. The LQG controller was designed in Simulink and subsequently autocoded using Simulink coder. In addition, updated input trim settings for all the control surfaces (estimated from Baldr flights 1 through 6) were used in this flight. This experiment used ONLY the left elevator of Baldr to regulate outputs around trim and track phi and theta