An Adaptive-Mesh Discretization Method for Sequential Convex Programming

Direct methods that use sequential convex programming (SCP) have gained popularity as computationally efficient and convergence-guaranteed algorithms for many space trajectory optimization problems. We developed a new SCP formulation that simultaneously optimizes the trajectory and the discretization mesh by including the time normalization parameters on each discretization segment as optimization variables. The proposed method enables mission designers to solve minimum-fuel problems with fewer discretization nodes while achieving exact bang-bang control profiles. It also allows to model variable time events such as free final time problems and planetary flybys.