Adaptive mascon modelling for small body gravity field reconstruction

The accurate reconstruction of the gravity field around irregular and heterogeneous small bodies is essential for safe and precise proximity operations in exploration missions. Mascon-based approaches provide a physically interpretable framework and can model heterogeneous mass distributions, but conventional implementations typically rely on static initialization, in which mascon points are predefined on uniform grids or polyhedral approximations and remain fixed during training, leading to inefficiency and limited adaptability in capturing complex boundaries or local variations. In this work, we propose an adaptive mascon framework that jointly optimizes mascon masses and their spatial configuration, and dynamically refines the mascon distribution in regions with high gravity field reconstruction error. Specifically, our method employs a 3D point-cloud–based initialization to accurately distinguish the asteroid boundary, a gradient-guided local refinement strategy to insert mascons where needed, and a position refinement to fine-tune mascon locations. Experiments on simulated and real asteroid data sets demonstrate that our approach achieves improved gravity reconstruction accuracy while using fewer parameters and maintaining moderate computational cost. This work offers a practical and efficient tool for enhancing navigation safety and mission planning in small body exploration.