Adaptive orbit determination for interplanetary spacecraft

The interplanetary orbit determination problem has been traditionally solved using least-squares techniques. Due to operational limitations of this method, a Kalman filter approach has been proposed for future missions. The proposed approach, known as the enhanced filter, includes all spacecraft and measurement modeling states in the filter. The goal of the enhanced filter is to increase the accuracy of the navigation process while utilizing only radiometric (Doppler and range) data. As an extension to the enhanced filter, an adaptive orbit determination approach (based on the Magill filter bank) has been developed here to process radiometric data. This adaptive approach can be used as a systematic method for the determination of the operational enhanced filter parameters, which are currently selected based on ad hoc methods. The first step in the development of the adaptive enhanced filter bank is the determination of the significant errors in the problem, which is accomplished using covariance analysis to develop an error budget. A high-fidelity simulation is used to generate tracking data and a bank of linearized Kalman filters is used to compute state estimates based on the tracking data. The Mars Pathfinder mission is utilized to demonstrate the effectiveness of the adaptive enhanced filter bank in determining variances for the process and measurement noise parameters based on the tracking data. For several important cases, the weighting factors for each filter in the bank are studied, along with the spacecraft state estimates at planetary encounter. To ensure that the single-run simulation results are not dependent on a particular realization of the various stochastic processes, Monte Carlo runs were made and average estimates are shown. The results for the range data case show that the adaptive enhanced filter bank is effective in selecting the process and measurement noise variances that match those used to generate the data. Results for the Doppler only case are not as conclusive, due primarily to linearization errors. In summary, an adaptive enhanced filter bank was developed and shown to effectively detect the proper process and measurement noise variances based on radiometric tracking data.