Learning-based collision-free coordination for a team of uncertain quadrotor UAVs

This paper investigates the safe coordination problem for a team of quadrotor unmanned air vehicles (UAVs) in the presence of model uncertainties and polygonal obstacles. Firstly, collision avoidance potential functions (PFs) related with both positions and velocities are utilized to construct the basic and bounded controller for achieving collision-free coordination. Then, in order to enhance the tracking performance under model uncertainties, an complementary control is developed via approximate dynamic programming (ADP), where the unknown dynamics of UAVs are accurately estimated in finite-time. Through using stored data in learning-based ADP, the approximated error of the weights in critic neural network converges with finitely excited signal. The sufficient conditions of the closed-loop system stability and collision avoidance are derived. The performance of the learning-based coordination methodology is demonstrated via simulation results. (C) 2021 Elsevier Masson SAS. All rights reserved.