An algorithm for robust initial orbit determination (IOD) under perturbed orbital dynamics is presented. By leveraging map inversion techniques defined in the algebra of Taylor polynomials, this tool is capable of not only returning an highly accurate solution to the IOD problem, but also estimating a range of validity for the aforementioned solution in which the true orbit state should lie. Automatic domain splitting is then used on top of the IOD routines to ensure the local truncation error introduced by a polynomial representation of the state estimate remains below a predefined threshold to meet the specified requirements in accuracy. The algorithm is adapted to three types of ground based sensors, namely range radars, Doppler-only radars and optical telescopes by taking into account their different constraints in terms of available measurements and sensor noise. Its improved performance with respect to a Keplerian based IOD solution is finally demonstrated with large scale numerical simulations over a subset of tracked objects in low Earth orbit.

翻译：本文提出了一种在摄动轨道动力学条件下进行稳健初始轨道确定的算法。通过利用泰勒多项式代数中定义的映射求逆技术，该工具不仅能够返回高度精确的初始轨道确定解，还能估计该解的有效范围——真实轨道状态应位于此范围内。随后，在初始轨道确定流程之上应用自动域分割技术，以确保状态估计的多项式表示所引入的局部截断误差保持在预设阈值以下，从而满足指定的精度要求。该算法针对三类地基传感器（即距离雷达、仅多普勒雷达和光学望远镜）进行了适配，考虑了它们在可用测量量和传感器噪声方面的不同约束。最后，通过对低地球轨道中部分被跟踪物体的大规模数值仿真，证明了该算法相较于基于开普勒模型的初始轨道确定解具有更优的性能。