Artifex Automata Predicting Robotic Chiseling Strokes

Two close-up views of textured stone surfaces showing different patterns of grooves and ridges. The top section has a rough, uneven texture with irregular, shallow indentations and small raised areas. The bottom section displays a smoother surface with longer, more linear grooves running mostly in one direction. The image highlights the contrast between these two types of stone textures, useful for comparing natural erosion or carving techniques.Historic buildings serve as vital links to our past, embodying environmental and cultural values, making their preservation and adaptive reuse a critical issue. However, the diminishing number of skilled craftspeople poses threats to these preservation efforts. This research explores the integration of robotics and machine learning with the traditional craft of stone chiselling, aiming to leverage digital technology for sustainable heritage conservation. A machine learning model trained on skilled craftsperson motion data coupled with a robotic arm demonstrates the ability to predict and execute chiselling strokes on sandstone. While surfaces produced by both the robotic system and the skilled artisan appear closely aligned, the presence of subtle differences challenge existing notions about craft and robotic fabrication. Beyond demonstrating the feasibility of integrating digital technologies into stonemasonry, this research promotes bridging the intricate realm of traditional craft with the integrative complexity of machine learning and the precision of robotics, and explores a paradigm shift in the way craft skills can be perceived and preserved.