Successive Convexification for Online Ascent Trajectory Optimization

Abstract

In this paper, a method based on the successive convexification is proposed to solve the ascent trajectory optimization problem, the algorithm converges to the optimal solution quickly even if the initial guess is coarse. A three-dimensional motion is formulated with complex aerodynamics and terminal constraints. Based on the modified aerodynamic coefficients, the new auxiliary control variables are designed to deal with the complex aerodynamics and non-smooth of control variables in the discrete optimization problem. The inner nonconvex constraints between the new control are relaxed to be convex without loss. The artificial infeasibility and unboundedness caused by linearization are tackled by the virtual controls and soft constraint for trust region in the successive convexification. The good convergence of the proposed method is illustrated by the iterative solutions of the ascent trajectory optimization problem for a small guided rocket, the accuracy is verified by the comparison with the optimal solution given by the typical optimal control solvers, and the feasibility and stability are demonstrated by optimal solutions of the ascent trajectory optimization problems under different missions and dispersed conditions. These excellent performances validated by the adequate simulations indicate that the proposed algorithm can be implemented online.

Keywords

• Convex optimization
• ascent trajectory optimization
• successive convexification
• online trajectory optimization
• complex nonlinear aerodynamic force