Robust To Acquire Landmark In The Process Of UAV Landing

Robust landmark acquisition is a key step in UAV landing process, in this paper, a robust elliptic fitting algorithm based on alternating directional multiplier method (ADMM) is proposed to obtain the landing landmark in the process of UAV landing. The main contributions of this paper are: ?) to enhance the robustness against outliers, we utilize the lp-norm into the direct least square fitting (DLSF) method. To solve the resultant non-convex and nonlinear optimization problem, auxiliary variables are introduced to decouple the elliptical parameters in the nonlinear objective function from the non-convex quadratic constraint. Then we determine the elliptical parameters and auxiliary variables alternately via ADMM; ?) to find the elliptical parameters, we develop the corresponding Lagrange multiplier method. In particular, the feasible region for the Lagrange multiplier is derived, which is efficiently computed by applying the bisection method; and iii) to calculate the auxiliary variables in parallel, we separate the corresponding objective function into multiple nonlinear sub-functions, where each of them is related to only one single auxiliary variable. Then, we analyze the convexity or concavity of each sub-function using its derivatives, which results in a simple but computationally attractive solver. Simulation and experimental examples are presented to demonstrate the robustness of the proposed ellipse fitting approach. Simulation and real data experiments demonstrate the robustness of the proposed method.