Design and Comparison of Control Schemes for UAV Autopilot

Abstract

Aves is an autopilot system developed primarily for fixed wing UAVs. The autopilot has been integrated with different aerial platforms varying in sizes as well as operational speeds. The control architecture of Aves is primarily PID with slight modification so as to include first order control and stability derivatives. The current paper presents the results of investigation of Non-linear Dynamic Inversion (NDI) and Linear Quadratic Regulator (LQR) control architecture, and comparison with existing control for full flight regime. NDI has the advantage of using the exact control requirements, provided the information about the dynamics is known. LQR has the advantage of Multi Input Multi Output (MIMO) system which can perform significantly better for transient as well as steady state in comparison to PID. The effects of digital implementation and sensor and process noise are taken into account in UAV simulation set up. Effects of uncertainty in stability and control derivatives on the performance are also presented.