ArduPilot-Based Adaptive Autopilot: Architecture and Software-in-the-Loop Experiments

This article presents an adaptive method for ArduPilot-based autopilots of fixed-wing unmanned aerial vehicles (UAVs). ArduPilot is a popular open-source vehicle software suite. We explore how to augment the PID loops embedded inside with model-free control method. The augmentation, adopted both attitude and total energy control, uses input/output data without requiring explicit model UAV. augmented architecture tested in software-in-the-loop UAV platform presence several uncertainties (unmodeled low-level dynamics, different payloads, time-varying wind, changing mass). performance measured terms tracking errors efforts loops. Extensive experiments original ArduPilot, proposed alternative autopilot strategies show that augmentation can significantly improve all payloads wind conditions: less affected by exhibits more than 70% improved tracking, 7% reduced effort.