Tiny Recursive Models for Solving the J2-Perturbed Lambert Problem

2026-05-30

AI Summary

This study addresses the computational inefficiency and reliance on manually designed homotopy paths in solving the J2-perturbed Lambert problem. To overcome these limitations, the authors propose the TRM-PL model, a recursive neural architecture with parameter sharing that performs end-to-end differentiable iterative optimization of the initial velocity through a two-level latent state representation, unifying initial guess generation and refinement. By replacing conventional homotopy strategies with a learning-driven recursive correction mechanism, the method achieves high-precision orbit solutions with only 2.3 million parameters, making it suitable for embedded deployment. Experimental results demonstrate significant accuracy improvements: terminal position errors in single-revolution LEO transfers decrease from 21.7 km to 0.027 km, and in multi-revolution LEO cases from 340.9 km to 0.31 km; with a single Newton correction, Jupiter orbit errors reach 0.063 km.

Abstract

This paper presents a fast, recursive neural solver for the J2-perturbed Lambert problem based on Tiny Recursive Models (TRM), termed the TRM-Perturbed Lambert (TRM-PL) model. TRM is a weight-shared architecture whose effective capacity emerges from iteration depth rather than parameter count: a compact reasoning module is applied repeatedly within a two-level latent hierarchy, refining a candidate departure velocity by simulating the J2 trajectory and correcting it from the resulting tracking error. This unifies initial-guess generation and iterative correction in a single, end-to-end differentiable architecture. The recursive refinement loop is a learned alternative to the homotopy and continuation schemes of classical perturbed-Lambert solvers: rather than following a hand-designed path from the Keplerian to the perturbed solution, the network learns its own sequence of corrections. We evaluate TRM-PL on three test cases of increasing difficulty: single-revolution low-Earth-orbit (LEO) transfers, multi-revolution LEO transfers, and multi-revolution Jovian transfers. Three training paradigms are compared: jointly learning the Lambert solution and the J2 correction; refining the Lambert initial velocity with target-position and J2-corrected velocity supervision; and refining it with target-position supervision alone. Across all cases, the refinement-only approaches are the most reliable. The position-supervised variant reduces the median terminal-position error from 21.7 km to 0.027 km on single-revolution LEO, from 340.9 km to 0.31 km on multi-revolution LEO, all with the same 2.3M-parameter architecture. A single Newton corrector iteration on the TRM-PL output tightens the Jovian median to 0.063 km, yielding compact models accurate enough for embedded deployment.

Problem

• J2-perturbed Lambert problem
• orbital transfer
• initial velocity determination
• trajectory correction
• spacecraft navigation

Innovation

• Tiny Recursive Models
• J2-perturbed Lambert problem
• recursive neural solver
• end-to-end differentiable architecture
• iterative refinement