Solving Optimal Control with Nonlinear Dynamics Using Sequential Convex Programming

Sequential convex programming (SCP) is a useful tool in obtaining real-time solutions to direct optimal control, but it is unable to adequately model nonlinear dynamics due to the linearization and discretization required. As nonlinear program solvers are not yet functioning in real-time, a tool is needed to bridge the gap between satisfying the nonlinear dynamics and completing execution fast enough to be useful. This paper presents a real-time control algorithm, sequential convex programming with nonlinear dynamics correction (SCPn), which ameliorates the performance of SCP under nonlinear dynamics. Simulations are presented to validate the efficacy of the method.