NMPC and genetic algorithm-based approach for trajectory tracking and collision avoidance of UAVs

摘要

Research on unmanned aircraft is improving constantly the autonomous flight capabilities of these vehicles in order to provide performance needed to employ them in even more complex tasks. UAV path planning (PP) system plans the best path to perform the mission and then it uploads this path on the flight management system (FMS) providing reference to the aircraft navigation. Tracking the path is the way to link kinematic references related to the desired aircraft positions with its dynamic behaviours, to generate the right command sequence. This paper presents a non-linear model predictive control (NMPC) system that tracks the reference path provided by PP and exploits a spherical camera model to avoid unpredicted obstacles along the path. The control system solves online (i.e., at each sampling time) a finite horizon (state horizon) open loop optimal control problem with a genetic algorithm. This algorithm finds the command sequence that minimises the tracking error with respect to the reference path, driving the aircraft far from sensed obstacles and towards the desired trajectory.