Evaluation and Analysis of ArduPilot Automatic Tuning Algorithm for the Roll Tracking Controller of a Small UAS

PID controllers are widely used for UAS flight control, and tuning of the control law is crucial to the success of a flight mission. In this paper, the automatic tuning process onboard the popular ArduPilot autopilot was examined, analyzed, and validated in both simulation and UAS flight test, considering both time domain and frequency domain controller specifications. The autotune algorithm was first implemented in MATLAB and simulated with a PID and PD controller using identified KHawk UAS lateral dynamic models. The results were further compared with fight test data. Frequency domain analysis showed the effectiveness of the autotune algorithm, as most of the autotuned controllers, except those with very aggressive gain sets, demonstrated performance comparable to controllers designed using a model-based approach. Specifically, the autotuned controllers showed high stability margins and desirable disturbance rejection capability.