The low cost of CubeSats is promising for deep space provided that the operations remain at low cost too. A realistic autonomous Guidance, Navigation and Control (GNC) would contribute. Here we present our road-map for BIRDYT, an autonomous GNC technology, and our latest developments regarding its In-Flight Orbit Determination (IFOD). The IFOD combines successive optical measurements of foreground objects, an asynchronous triangulation and a Kalman filter. The measurements are expected from an object tracker function that is presented here. Then the triangulation is presented, based on a weighted least-squares method. The solution is still noisy and feeds a Kalman filter in order to get improved. The performance is assessed for a case study in cruise context, an interplanetary journey from Earth to Mars. This case study is a necessary step before adapting to proximity operations at an asteroid. A software architecture for Monte-Carlo simulations is implemented to embed the IFOD algorithm and to assess how the optical errors propagate. The latest results are presented. A full interpretation would require further mathematical analysis. As of now, the accuracy reached in cruise context is better than 100 km transversely to the motion at 1-σ in a conservative case and some improvements are still possible. This autonomous orbit determination could be adapted to further deep-space contexts, with a priority for an auxiliary probe of a mothercraft exploring an asteroid.