ESA GNC Conference Papers Repository

This article describes the design and validation of an autonomous Guidance Navigation and Control (GNC) for Phobos Sample Return mission for the Descent and Landing phase. It outlines the main drivers for the design choices including operational constraints, pinpoint landing requirements, environmental conditions: Sun exposition and day duration on Phobos, complex three body dynamics, initial knowledge and dispersions. An extensive mission analysis is first presented in order to support the descent and landing strategies trade-off for a pinpoint landing on Phobos. The specific and challenging environment is addressed in term of complex dynamics with a three body problem, illumination conditions required for the visual based navigation during the descent and landing, and stringent pinpoint landing requirements (50m accuracy). This article highlights as well the operational constraints and limitations like the short timeline for on-ground processing and uplink telecommands and their implication on the strategy choices. Based on the mission analysis and strategies trade-off performed, two strategies for the D&L have been chosen and described. The strategies differ by their required level of autonomy in the GNC: - Direct descent, in which a transfer to gate (ground-commanded) is followed by a closed loop, autonomous D&L with vision based navigation. - Descent with Hovering, in which a closed loop station-keeping is performed in order to communicate with the ground for GO/NOGO command. A full 6 degree of freedom GNC - translation and attitude - has been designed for these two strategies including: mode management, guidance (with offline trajectory optimisation), navigation based on tracked features by the image processing, and continuous control with RCS. An enhanced relative navigation has been developed in order to provide position estimate with respect to the landing site based on unknown features detection and mapping with on-board stored database. The database generation procedure is explained with an overview of the operational constraints derived from it: to generate the sufficient number of images with different viewpoints to allow robust features matching and position estimation. The validation campaign results of the designed GNC SW is discussed. Monte-Carlo simulations, sensitivity and robustness analyses in Model in the loop and software in the loop are reported. Two levels of fidelity simulators: a medium fidelity with a performance model of the image processing chain, and a high fidelity including a scene generator and image processing, have been designed for the testing. A real time breadboard of the GNC system has been implemented with representative HW for space (LEON-2 for the GNC, FPGA Virtex5 for the image processing), representative interface with the images scene generator through Spacewire. The resultant GNC solution proposed constitutes a solid basis for future GNC design for Phobos Sample Return mission and for future exploration missions for which similar GNC technologies can be used.