Statistical Entry, Descent, and Landing Flight Reconstruction with Flush Air Data System Observations using Inertial Navigation and Monte Carlo Techniques

A method is introduced to consider flush air data system (FADS) pressures using a technique based on inertial navigation to reconstruct the trajectory of an atmospheric entry vehicle. The approach augments the recently-developed Inertial Navigation Statistical Trajectory and Atmosphere Reconstruction (INSTAR), which is an extension of inertial navigation that provides statistical uncertainties by utilizing Monte Carlo dispersion techniques and is an alternative to traditional statistical approaches to entry, descent, and landing trajectory and atmosphere reconstruction. The method is demonstrated using flight data from the Mars Science Laboratory (MSL) entry vehicle, which contained an inertial measurement unit and a flush air data system called the Mars Entry Atmospheric Data System (MEADS). An MSL trajectory and atmosphere solution that was updated using landing site location in INSTAR is first presented. This solution and corresponding uncertainties, which were obtained from Monte Carlo dispersions, are then used in a minimum variance algorithm to obtain aerodynamic estimates and uncertainties from the MEADS observations. MEADS-derived axial force coefficient and freestream density estimates and uncertainties are also derived from the minimum variance solutions independent of the axial force coefficients derived from computation fluid dynamics (CFD), which have relatively high a priori uncertainty. Results from probabilistic analyses of the solutions are also presented. This dissertation also introduces a method to consider correlated CFD uncertainties in INSTAR. From a priori CFD uncertainties, CFD force and pressure coefficients are dispersed in a Monte Carlo sense and carried over into the reconstructions. An analysis of the subsequent effects on the trajectory, atmosphere, and aerodynamic estimates and statistics is presented. Trajectory, atmospheric, and aerodynamic estimates compare favorably to extended Kalman filter solutions obtained by the MSL reconstruction team at NASA Langley Research Center. The uncertainties obtained through the methods from this work are generally smaller in magnitude because of assumptions made regarding sources of error in the MEADS pressure transducer uncertainties. Using data-derived uncertainties in the pressure measurement noise covariance results in aerodynamic parameter estimate uncertainties that are in better agreement with the uncertainties derived from the Monte Carlo dispersions. CFD database errors dominate the uncertainties of parameters derived from aerodatabase axial force coefficients