1: \begin{abstract}

2: The indirect method has long been favored for solving trajectory optimization problems

3: due to its ability to reveal the structure of the corresponding optimal control. However, providing an appropriate initial guess of the co-state vector remains challenging due to its lack of physical significance. To address this issue, this paper introduces a Physics-Informed Indirect Method (PIIM), \textcolor{blue}{for which all shooting variables are constrained into a narrow space by utilizing their physics information.}

4: Starting from the Time-Optimal Soft Landing Problem (TOSLP), an analytical estimation method for the minimum flight time is provided; we show that the co-state vector at the final time can be constrained on a unit 3-D hypersphere by eliminating the mass co-state and the numerical factor used in the initial co-state vector normalization; and the physical significance of the optimal control at the final time is exploited, allowing to further narrow down the solution space for the co-state vector to a unit 3-D octant sphere. Then, by incorporating the above physics-informed information into a shooting function that involves of propagating the dynamics of both states and co-states backward, the PIIM achieves faster and more robust convergence for the TOSLP compared to existing indirect methods.  In addition, the combination of the PIIM with a homotopy approach is proposed, allowing one to solve the \textcolor{blue}{fuel-optimal soft landing problem robustly. Finally, numerical simulations are presented to demonstrate and validate the developments of the paper.}

5: % demonstrate that,  while the success rate of the conventional indirect method for solving the FOSLP is $89.35\%$,  the proposed method takes a shorter time to find the solution to the FOSLP with a success rate of up to $100\%$.}

6: \end{abstract}

7: