A polynomial-based constrained solver for fuel-optimal low-thrust trajectory optimization

Abstract: Differential Dynamic Programming (DDP) is a trajectory optimization method, particularly resilient to poor initial guesses. However, its long run times compared to other methods make it less suitable for embedded systems. In this work, we introduce polynomial-based DDP methods capable of enforcing constraints while optimizing for fuel efficiency. Additionally, a polynomial-based Newton solver is implemented to enforce constraints with high precision. The proposed solver, Differential Algebra-based Differential Dynamic programming (DADDy), is validated and tested on various astrodynamics scenarios. Results demonstrate that DADDy achieves the same solutions as state-of-the-art DDP methods but with significantly reduced run times. Specifically, for the scenarios investigated in this work, the most stable method was able to achieve 100% convergence and achieved runtime reductions of 70% in the Sun-centered two-body problem, 23% to 94% in the Earth-Moon CR3BP, and 46\% to 59% in the Earth-centered two-body problem.