UAV Flight Control System Design
The Viking Aerospace approach to flight control system design consists of four phases: Modeling, Multivariable Controller Synthesis, Controller Simulation, and Flight Validation. Each of these phases can consist of several different methods depending on the exact application. No matter which methods are used, much effort has been expended to automate as much of the process as possible leading to very rapid controller design implementation. Typically, we are able to generate a dynamics model, synthesize a set of controllers, perform controller simulations, and conduct a flight validation in as little as 4 hours.
We at Viking Aerospace pride ourselves on being able to rapidly generate accurate flight dynamcis models of vehicles. In the dynamic modeling world there is a tradeoff between accuracy and costs/time associated with achieving that accuracy. We try to maintain the perfect balance between the two goals by combining multiple modeling methods with a high amount of automation to expedite the process as much as possible. The primary modeling method we use is multivariable system identification through flight testing. Essentially we perform several flights with specific maneuvers that excite the various modes of the vehicle. We then utilize our proprietary software to generate flight dynamics models of the vehicle using a combination of time and frequency domain methods. The modeling phase is fundamental to the final product because the performance of the flight control system will be directly affected quality of the dynamics model. However, there are always discrepencies between the linear flight dynamics model, and the real world vehicle, which is definitely non-linear. For this reason, we perform a great deal of sensitivity studies around the base dynamics model to account for any discrepencies. Again, this process is highly automated to decrease both the time and cost associated with this process.
Multivariable Controller Synthesis
Viking Aerospace utilizes the wePilot flight control system for all helicopter and fixed wing unmanned aircraft. Viking Aerospace synthesizes multivariable controllers based on h∞ methods. The primary control problem associated with small helicopters is that there is a significant amount of coupling between the longitudinal and lateral modes of the vehicle. This coupling calls for the need to synthesize controllers in a multivariable sense, simultaneously handling both the longitudinal and lateral modes, instead of the classical approach of attacking these modes independently in a single-input, single-output design. The use of h∞ controller methodologies results in a perfect balance between controller performance and robustness. The term robustness has a clear definition in the controls world. It means that the controller is guaranteed to be stable even if the dynamics model changes from the nominal state. The h∞ controller design process focuses on optimizing performance with a certain level of robustness. This results in vehicles that have excellent controller performance (station keeping in gusty winds), but can handle relatively wide changes in flight dynamics (different payloads, center of gravity, etc.).
Once a controller has been synthesized, it is run through various simulations to verify disturbance rejection, sensitivity to noise, as well as sensitivity to changes in the vehicle dynamics. At this point in the design phase we create various modifications of the base vehicle dynamics model and simulate each one with the synthesized controller to verify that the controller remains stable in the presence of changes to the vehicle dynamics. Essentially, we are verifying the robustness of the controller that was synthesized. As with the modelling, much of this process has been automated to decrease the time required to validate the controller.
The final step in the controller design process is to validate the controller through flight testing. Viking runs the controller through various preprogrammed missions to verify controller performance. The flight verification consists of hover testing in various wind conditions, high speed maneuvers in all directions, and waypoint tracking accuracy verification. Our systems achieve hover and flight path tracking of less than 1 meter.