L1 Adaptive Backstepping for Robust Trajectory Tracking of UAVs

This paper addresses the trajectory tracking control (i.e., outer-loop control) problem for unmanned aerial vehicles (UAVs) in the presence of modeling uncertainties and external disturbances. The kinematics and dynamics of the trajectory tracking problem are always in strict feedback form. While there is no uncertainty in the kinematics, rapidly changing uncertainties present in the dynamics makes this problem an ideal candidate for L-1 adaptive backstepping control. The trajectory tracking controller serves as the outer-loop in a cascaded design architecture and supplies a reference quaternion and thrust to the inner-loop controller. The inner-loop controller in turn generates a moment demand for the control allocation module. Such an inner-outer loop architecture is modular and does away with the requirement that the commanded trajectory be four times differentiable as would be required if a single monolithic backstepping control law where to be used. Both simulations and flight tests are used to demonstrate the effectiveness of the proposed controller and how the controller tracks any twice differentiable trajectory, respectively.