Computational design for lunar infrastructure built with unprocessed stones

Lunar infrastructure construction requires innovative strategies to minimize energy consumption and human intervention. This study presents a computational design method for robotic construction of lunar masonry structures using in-situ, unprocessed stones. The method iteratively determines the optimal placement of stones through an optimization formulation that incorporates both geometric and physical constraints. To achieve computational efficiency, the stones and the target structure are encoded in 3D tensors, and their geometric features are evaluated using discrete convolution. Stability of stone placement is assessed both geometrically, as optimization constraints, and numerically, through the Non Smooth Contact Dynamics method (NSCD). The proposed computational design method is applied in the planning of arches, domes, and walls, showing versatility across building components while also identifying limitations on the geometry of spanning elements. The robotic construction experiment for building a wall composed of 138 stones achieves a placement success rate of 95%, validating the construction feasibility and energy efficiency of building lunar infrastructure using unprocessed stones.