ESA GNC Conference Papers Repository

Accurate autonomous navigation capabilities are essential for future lunar robotic landing missions with a pin-point landing requirement, since in the absence of direct line of sight to ground control during critical approach and landing phases, sufficient navigation accuracy is needed to establish a guidance solution to reach the landing site reliably. The purpose of the DLR-internal ATON project series (Autonomous Terrain-based Optical Navigation) is the conception and implementation of a complete integrated navigation system for such an autonomous lunar pin-point landing scenario.The system fuses inertial with relative and absolute optical measurements of spacecraft attitude and position, providing a navigation solution that enables the lander to satisfy navigation accuracy requirements throughout all mission phases from descent orbit injection until touchdown. Subject of this paper is the processing of data collected from flight tests that consisted of scaled descent scenarios where the unmanned helicopter of approximately 85 kg approached a landing site from altitudes of 50 m down to 5 m for a downrange distance of 150 m. In this test environment, the navigation hardware setup had to work under effects such as vibration, camera distortion, vehicle disturbances, etc. Printed crater targets were distributed along the ground track and their detection provided earth-fixed measurements. The algorithm used to detect and match the crater targets is an unmodified method used for real lunar imagery. We show navigation solutions derived from data recorded during these flight tests of the assembled ATON hardware demonstrator. We investigate the attainable quality of the vehicle pose estimate using a minimum set of measurements: vehicle accelerations and angular rates as provided by a tactical-grade Inertial Measurement Unit (IMU), and absolute measurements of vehicle position and attitude as provided by a crater detection and matching algorithm.The high-rate inertial propagation is corrected and calibrated with the less frequent crater navigation fixes through a closed-loop, loosely coupled hybrid navigation set-up. An indirect filtering scheme is used, accounting for and estimating the most influential IMU error sources. Correct modelling and tuning of the navigation algorithm as well as consistency of inertial and visual sensor data are demonstrated through the analysis of filter innovation statistics.Predicted attainable accuracy of the fused solution is also supported by comparison with the on-board solution of a dual-antenna high-grade GNSS receiver.