Orbit Determination of Mixed Observations of Multiple Objects1,2

In the conventional orbit determination with optical measurements of space objects, some observations of different objects may be marked as the same object. For this kind of data, the process of orbit determination will be failed to converge or completely erroneous because of the composition of multiple objects. Here, a method is proposed from the viewing point of maximum likelihood, which combines the orbit improvement and object identification by employing the EM (Expectation Maximum) method, and in the implementation of this method, a robust estimation will be given. Corresponding numerical simulations indicate that this method is feasible, effective, and convenient.