Abstract: A model compensation version of successive convex programming is developed to account for aerodynamic dispersions and applied to the atmospheric rocket trajectory planning problem. A strategy to successively compensate the nonlinear terms in the vehicle dynamics is proposed, with which the original problem can be transformed into a convex one and then solved rapidly. A method to successively convexify the path constraints is developed, which is capable of addressing the nonconvex path constraints without slowing the optimization process. At last, the convergence is analyzed theoretically. The simulation results suggest that the proposed method can generate trajectories autonomously and rapidly for launch vehicles both in landing phase and in ascent flight. Besides, the convergence rate of the method meets the real time requirement.

Key words: Launch vehicles, Successive convex programming, Online trajectory planning, Endoatmospheric, Model compensation