Reference command shaping for approximate dynamic inversion based model Reference Adaptive Control

A hybrid adaptive-optimal control architecture is presented, which is suitable for implementation on systems with fast, nonlinear and uncertain dynamics subject to constraints. Approximate dynamic inversion transforms the nonlinear system into an equivalent linear form. A linear error controller allows a designer chosen reference model to be tracked by the inverted plant. Model Reference Adaptive Control uses a weighted combination of basis functions in order to reduce the impact of modeling uncertainties on the closed loop dynamics. The underlying assumption is, that if the uncertainty is cancelled, the plant behaves like the inversion model and tracks the reference model exactly. Reference models are often chosen to be conservative. By employing a Model Predictive Controller to shape the reference command, the full control authority of the plant can be exploited, while simultaneously abiding input and state constraints. Since the reference model is preselected, the optimal control problem can be solved offline, drastically reducing the onboard required computational power. The control architecture is implemented for the attitude loop of a multirotor system and validated in flight test.