Online corrections to neural policy guidance for pinpoint powered descent

This study presents incremental correction methods for refining neural network parameters or control functions entering into a continuous-time dynamic system to achieve improved solution accuracy in satisfying the interim point constraints placed on the performance output variables. The proposed approach is to linearize the dynamics around the baseline values of its arguments and then to solve for the corrective input required to transfer the perturbed trajectory to precisely known or desired values at specific time points, in other words, the interim points. Depending on the type of decision variables to adjust, parameter correction and control function correction methods are developed. These incremental correction methods can be used as a means to compensate for the prediction errors of pretrained neural networks in real-time applications where high accuracy of the prediction of dynamical systems at prescribed time points is imperative. In this regard, the online update approach can be useful for enhancing overall targeting accuracy of finite-horizon control subject to point constraints using a neural policy. A numerical example demonstrates the effectiveness of the proposed approach in an application to a powered descent problem on Mars.