Minimum Snap Trajectory Tracking for a Quadrotor UAV using Nonlinear Model Predictive Control

We present an indoor autonomous quadrotor flight that incorporates motion planning, trajectory generation, and trajectory tracking. The asymptotically optimal Rapidly-exploring Random Tree* (RRT*) algorithm is used to generate a set of obstacle-free waypoints. In highly cluttered settings, effective deviations of the attitude should be permitted allowing a greater range for roll and pitch angles hence an exact nonlinear model was derived using Newton and Euler formulations. The minimum snap cubic spline algorithm is used to generate a dynamically feasible optimal trajectory passing through the waypoints then a nonlinear model predictive control (NMPC) is implemented to track this trajectory. Simulations are carried out in both two and three-dimensional obstacle cluttered environments and the results are discussed.