NMPC Strategy for a Quadrotor UAV in a 3D Unknown Environment

This work presents a Nonlinear Model Predictive Control strategy for mobile robot navigation in unknown environments. The control system aims to reach a goal safely, as fast as possible, minimizing the control effort, and the distance between the current trajectory and the goal. A LIDAR (Light Detection and Ranging) sensor is used to obtain obstacles information as these are approached by the robot. The LIDAR output is then processed to obtain inequality constraints describing a collision-free area. This area is partitioned in convex subregions in order to be used within the proposed NMPC approach to perform path tracking, ensuring obstacle avoidance. Numerical results are provided to corroborate the performance of the system.