Trajectory Optimization of Constrained UAV Problem Based on Gauss Pseudospectral Method and Multiple Shooting

Trajectory optimization of unmanned aerial vehicle (UAV) is one of the frontiers and hotspots of UAV researches. It is usually modeled with various constraints generated by physical limitations, safety requirements and obstacles in the environment. These simultaneous constraints improve the difficulties to solve the trajectory optimization problem. An effective method based on Gauss pseudospectral method and multiple shooting is thereby proposed, where the control vector, the initial state on the segments, and the terminal time are parameterized and treated as independent variables, the state at collocation points are param-eterized as dependent variables in the nonlinear programming problem (NLP). One advantage of this parameterization strategy is that the constraints can be handled directly. To accelerate the optimization process and improve the efficiency, the sensitivities of state at the collocation points with respect to independent variables are derived. The proposed optimization structure and detailed algorithm are given. The simulation results illustrate the effectiveness of proposed method on constraints processing and obstacles avoidance.