Search papers, labs, and topics across Lattice.
This paper trains neural network surrogates to approximate the cost and feasibility of low-thrust trajectories, which are traditionally computationally expensive to evaluate. They find that performance scales linearly with the logarithm of training data and network parameters, and introduce a self-similar transformation to enable generalization across different orbital environments. The resulting models accurately predict fuel consumption and transfer time, and are validated on several benchmark problems.
One neural network can now rapidly estimate optimal fuel consumption and transfer time for low-thrust trajectories across diverse orbital environments, thanks to a novel self-similar transformation.
Low-thrust trajectory design relies heavily on repeated evaluations of fuel consumption and transfer feasibility, which require expensive optimal control solutions. In this work, we show these quantities can be accurately approximated by machine learning surrogates, enabling fast and scalable evaluation across a wide range of scenarios. By increasing both dataset size and model capacity, we observe that low-thrust trajectory optimization follows a scaling law, with performance improving linearly with the logarithm of training data and network parameters, and no evidence of saturation within the explored regime. Guided by this observation, we construct a large-scale dataset using the proposed homotopy-ray strategy tailored to mission design requirements. A key is the introduction of a self-similar transformation, which allows generalization across semi-major axes, inclinations, and central bodies avoiding retraining. As a result, the same neural approximator can be applied to diverse orbital environments and mission classes. The proposed models accurately predict optimal fuel consumption and minimum transfer time for single- and multi-revolution transfers. Their performance and generalization are demonstrated on a public dataset, a multi-asteroid flyby problem from the Global Trajectory Optimization Competition, and an asteroid rendezvous mission design. The models and datasets are released as open-source to support the space community.
Tsinghua AI