Robust Low-Thrust Trajectory Design for Interplanetary Spaceflight: An Adaptive Latent Reinforcement Learning Method

TY - JOUR

T1 - Robust Low-Thrust Trajectory Design for Interplanetary Spaceflight

T2 - An Adaptive Latent Reinforcement Learning Method

AU - Gao, Han

AU - Lin, Yanghui

AU - Sun, Zhongqi

AU - Cui, Bing

AU - Zhang, Guangchen

AU - Xia, Yuanqing

N1 - Publisher Copyright: © 2013 IEEE.

PY - 2026/6/1

Y1 - 2026/6/1

N2 - This article investigates the problem of robust trajectory design for low-thrust spacecraft subject to state and observation uncertainties. An adaptive latent reinforcement learning (RL) scheme based on sequential latent variable models (SLVMs) is proposed to address this issue. First, an SLVM is employed for the representation learning of uncertain environments and for predicting future observations. Subsequently, by integrating representation learning based on the SLVM with proximal policy optimization (PPO), a stochastic latent PPO (SLPPO) scheme is introduced. Distinct from existing methods, the control policy is derived from learned stochastic latent variables rather than raw uncertain observations, which effectively mitigates the adverse impact of uncertainties on control performance. Furthermore, to enhance training efficiency, an improved dense reward shaping mechanism is designed based on the observation predictions from the SLVM and adaptive techniques. Finally, numerical simulations of two rendezvous missions validate the effectiveness of the proposed approach.

AB - This article investigates the problem of robust trajectory design for low-thrust spacecraft subject to state and observation uncertainties. An adaptive latent reinforcement learning (RL) scheme based on sequential latent variable models (SLVMs) is proposed to address this issue. First, an SLVM is employed for the representation learning of uncertain environments and for predicting future observations. Subsequently, by integrating representation learning based on the SLVM with proximal policy optimization (PPO), a stochastic latent PPO (SLPPO) scheme is introduced. Distinct from existing methods, the control policy is derived from learned stochastic latent variables rather than raw uncertain observations, which effectively mitigates the adverse impact of uncertainties on control performance. Furthermore, to enhance training efficiency, an improved dense reward shaping mechanism is designed based on the observation predictions from the SLVM and adaptive techniques. Finally, numerical simulations of two rendezvous missions validate the effectiveness of the proposed approach.

KW - Latent variable model

KW - reinforcement learning (RL)

KW - reward function

KW - robust trajectory optimization

UR - https://www.scopus.com/pages/publications/105027462092

U2 - 10.1109/TCYB.2026.3651240

DO - 10.1109/TCYB.2026.3651240

M3 - Article

AN - SCOPUS:105027462092

SN - 2168-2267

VL - 56

SP - 3249

EP - 3262

JO - IEEE Transactions on Cybernetics

JF - IEEE Transactions on Cybernetics

IS - 6

ER -