ESA GNC Conference Papers Repository

Inertial-only navigation is unable to reach the 200-m (3 ) landing accuracy required by future lunar missions. It needs to be augmented by vision-based measurements. The techniques currently investigated consist in detecting features in images of the planet surface and matching them into a geo-referenced database. This information, fused with the inertial measurements using an Extended Kalman Filter (EKF), leads to accurate position estimation. The presence of relatively well-shaped crater impact covering the surface of several celestial bodies makes the crater detection and matching algorithm an attractive solution for future autonomous navigation systems. This paper proposes an innovative, robust and accurate approach to perform the detection and matching of craters. Using the lunar landing mission as a reference, the algorithm has been extensively validated: with real images from previous mission, with synthetic images under various camera orientation and lighting conditions, with end-to-end, closed-loop and high-fidelity simulations, with a real-time test bench equipped with a flight-like camera and a lunar surface mock-up.