Constrained Control for Autonomous Spacecraft Rendezvous: Learning-Based Time Shift Governor

TL;DR

The paper tackles constrained autonomous spacecraft rendezvous by augmenting a nominal discrete-time LQ controller with a Time Shift Governor (TSG) that uses a time-shifted Chief trajectory as reference. It introduces a learning-based variant (L-TSG) that employs an LSTM to predict the time shift from state sequences, enhanced with a phase-adaptive sliding window and a custom loss, plus a hybrid verification mechanism that can fallback to conventional TSG. Simulation results in both LEO and Molniya orbits show substantial reductions in online computation time (from around 0.15 s to ~0.02 s) and robust constraint enforcement (LoS, thrust, and approach-velocity) while ensuring rendezvous within the desired reference. The approach demonstrates that combining imitation-learning-based time-shift estimation with a safety verifier yields fast, reliable RD performance in the Two-Body setting, with potential applicability to more complex dynamics. Overall, L-TSG provides a practical, computation-efficient pathway to autonomous RD under multiple state and control constraints.

Abstract

This paper develops a Time Shift Governor (TSG)-based control scheme to enforce constraints during rendezvous and docking (RD) missions in the setting of the Two-Body problem. As an add-on scheme to the nominal closed-loop system, the TSG generates a time-shifted Chief spacecraft trajectory as a target reference for the Deputy spacecraft. This modification of the commanded reference trajectory ensures that constraints are enforced while the time shift is reduced to zero to effect the rendezvous. Our approach to TSG implementation integrates an LSTM neural network which approximates the time shift parameter as a function of a sequence of past Deputy and Chief spacecraft states. This LSTM neural network is trained offline from simulation data. We report simulation results for RD missions in the Low Earth Orbit (LEO) and on the Molniya orbit to demonstrate the effectiveness of the proposed control scheme. The proposed scheme reduces the time to compute the time shift parameter in most of the scenarios and successfully completes rendezvous missions.

Figures (12)

• Figure 1: ECI and VNB coordinate systems.
• Figure 2: Diagram of the control system architecture incorporating learning-based predictive model with TSG and DTLQ controller.
• Figure 3: Diagram of the LSTM cell, illustrating the input, forget, and output gates.
• Figure 4: Overview of LSTM-based time shift prediction model network architecture.
• Figure 5: Trajectories expressed in the VNB frame: the Deputy spacecraft’s path (gray line) tracks the virtual target’s trajectory (magenta asterisk).