A Symplectic Orbital Estimator for Direct Tracking on Satellites

In this paper we describe a new method useful for autonomous onboard estimation for satellites in LEO. The method employs symplectic integration which provides very high modelling precision in sufficiently short times to be computed by onboard computers. We demonstrate a modelling accuracy of less than 1 centimeter in position after 7 days propagation of LEO orbits by filtering 700 simulated noise-free position-velocity measurements in 1 minute CPU time on a Pentium 200 MHz processor.We demonstrate the accuracy of the filter when noise is added to simulated data and argue that sparse measurements are more suited to this filter. For the purposes of evaluating this method we have ignored the effects of atmospheric drag and radiation pressure, although we describe briefly how the method can be modified to include these effects for real applications.