Velocimeter-Aided Attitude Estimation for Mars Autonomous Landing: Observability Analysis and Filter Algorithms

Abstract

This articlefocuses on the attitude estimation problem for a Mars landing that employs an inertial measurement unit and a velocimeter. At the beginning of the parachute descent phase, high dynamic oscillatory motion may degrade the gyroscope performance or saturate the gyroscope, thereby producing large attitude estimation errors and a high landing risk. To address this problem, an attitude estimation algorithm with nadir vector correction is proposed by combining data from the IMU and the velocimeter. First, the body-frame nadir vector is determined from the gravitational acceleration, which is estimated in conjunction with the body velocity aided by velocimeter measurements in the framework of the Kalman filter. The observability analysis shows that the nadir vector may be estimated as long as two velocimeter beams are available. Next, to mitigate the large attitude knowledge errors caused by high dynamic motion, a purely deterministic approach is used to combine the estimated nadir vector with the inertial navigation system propagated attitude such that the attitude knowledge error along the nadir vector direction can be reduced. Simulation results demonstrated that the proposed algorithms can cope with gyroscope performance degradation or gyroscope saturation, enabling the embedded autonomy of lander systems.