Nonlinear MPC for Multiple Quadrotors in Dynamic Environments

Abstract

Accurate trajectory tracking for multiple quadrotors is essential for safe navigation in dynamic environments. However, this can be a challenge due to high nonlinearities of the systems, environment's obstacles and constraints in control actions. This article presents a nonlinear model predictive control (NMPC) for tracking a variety of trajectories using multiple quadrotors in a unknown environment with dynamic and static obstacles. The controller is formulated using kinematic and dynamic models of multiple commercial aerial robots (Mavic Pro 2), actuators limitations, and finally converting obstacles objects to states constraints. It was developed through CasADI, an open-source software for numerical optimization that provides a fast solution to keep each quadrotor on the desired trajectory while nonlinear constraints guarantee collision avoidance and smooth control signals. The controller is implemented in simulations to systematically evaluate the performance, considering tracking accuracy and computational time. The results of the simulations are presented to confirm that the proposed NMPC generates a smooth control values ensuring safe states for each aerial robot while following a desired trajectory in unknown environment.